MBA Thesis

Impact of Artificial
Intelligence on Management
and Leadership in Research
& Development
A Case Study of Thermo Fisher Scientific

Somar Al-Walai, Jianwei Liang

Supervisor
Anupama Unnikrishnan

Karlskrona, Sweden
June 2021

www.bth.se/mba DEPARTMENT OF INDUSTRIAL ECONOMICS

This thesis is submitted to the Department of Industrial Economics at Blekinge Institute of
Technology in partial fulfilment of the requirements for the Degree of Master of Science in Industrial
Economics and Management. The thesis is awarded 15 ECTS credits.

The author(s) declare(s) that they have completed the thesis work independently. All external
sources are cited and listed under the References section. The thesis work has not been submitted in
the same or similar form to any other institution(s) as part of another examination or degree.

Author information:
Somar Al-Walai
somarwalai@gmail.com

Jianwei Liang
liangjianwei2010@gmail.com

Department of Industrial Economics

Blekinge Institute of Technology
SE-371 79 Karlskrona, Sweden

Website: www.bth.se
Telephone: +46 455 38 50 00
Fax: +46 455 38 50 57
Abstract
Background: In current business world, big data and Artificial Intelligence (AI) implementation in daily
business has become a mega-trend in different organization functions across various industries. The
questions of what impacts against management and leadership can be expected with AI application, and
how to make the most of AI application to achieve innovation success have become extremely
interesting for business leaders, especially in R&D department which is considered to be innovation
locomotive of a company.
Objectives: Existing studies on implementation of AI application in research and development (R&D)
department is rather scarce although there is a strong relevance for AI to be brought into R&D for
achieving successful innovation. This study tries to fill this gap in literature aiming to explore the impacts
of AI application on management and leadership, and also how AI can be used for achieving innovation
success in R&D department.
Method: The research methodology chosen for this study is a single and holistic case study on the
organisation Thermo Fisher Scientific as we considered our research question to be unexplored and rare.
We have used multiple sources of data including both primary data from multiple interviews and
secondary data in the form of publicly available data. For the analysis of the data, we have used the
grounded theory approach with 9 interviews for a phenomenological study.
Results: A theoretical dynamic model was created for explaining the mechanism of AI influences in R&D
by using grounded theory based on empirical interview data analysis. Interesting findings shed light on
importance of implementation of AI application in R&D. The results disclose that AI application in R&D
can lead to higher efficiencies in processes, decision making, costs and stimulate innovation, while a
shift of leadership elements and organizational structure changes can be expected.

Keywords:
Artificial Intelligence, leadership, management, R&D, innovation

i
Acknowledgement:
To complete a case study consumes considerable time and efforts. However, without the generous and
kind support from other people, this work would have not been possible to proceed and finalize despite
how much time and efforts spent. Therefore, we would like to take the opportunity to express our
sincere gratefulness to all the interviewees who were willing to spend the time in the interviews and
generously share their valuable thoughts, interesting examples and opinions with us, which is with not
doubt the crucial foundation for the study. We have been impressed by the strong openness and
curiosity of all the participants and this has strengthened our confidence in continue studying in the
research topic. Meanwhile, we are grateful to have received the valuable positive and critical comments
from our peer students in the opposition reports and presentation. Last but not least, we would like to
send our special thanks to our tutor Anupama Unnikrishnan who has been always highly supportive in
giving us constructive guidance and comments from Day 1 to the completion of the study. Her abundant
experience and continuous support have not only enabled us to learn much in thesis writing but also
encouraged us to continuously improve our work.

June 2021
Somar Al-Walai & Jianwei Liang

ii
Table of Contents

1. Introduction..................................................................................................................... 1
1.1. Background ............................................................................................................................................................ 1
1.2. Problem Formulation .......................................................................................................................................... 2
1.3. Definitions.............................................................................................................................................................. 4
1.3.1. Artificial Intelligence in a Nutshell ........................................................................................ 4
1.3.2. Transformational leadership ................................................................................................. 6
1.4. Delimitations ......................................................................................................................................................... 7
1.5. Structure ................................................................................................................................................................ 8

2. Literature Review ........................................................................................................... 9

2.1. The Impacts of AI Application on Leadership and Employees in Business ............................................ 9
2.2. How AI Can be Best Utilized by Leaders to Achieve Innovation Success in R&D ............................ 10
2.3. Theoretical Framework .................................................................................................................................... 10

3. Methodology ................................................................................................................. 14
3.1. Selection of Research Approach .................................................................................................................... 14
3.2. Research Design ................................................................................................................................................. 15
3.3. Choosing Target Company for Case Study ................................................................................................. 16
3.3.1. Thermo Fisher Scientific as a Company .............................................................................. 16
3.3.2. R&D in Thermo Fisher Scientific.......................................................................................... 17
3.3.3. AI Development in Thermo Fisher Scientific ....................................................................... 17
3.4. Data Collection .................................................................................................................................................. 18
3.4.1. Sources of Data ................................................................................................................... 18
3.4.2. Sample Selection ................................................................................................................. 19
3.4.3. Basic Information of the Interviewees ................................................................................ 19
3.4.4. Interview Design ................................................................................................................. 20
3.5. Data Analysis ....................................................................................................................................................... 21
3.6. Ethical Considerations ...................................................................................................................................... 23
3.7. Validity and Reliability ....................................................................................................................................... 23
3.7.1. Construct Validity ................................................................................................................ 23
3.7.2. Internal Validity ................................................................................................................... 24

iii
 3.7.3. External Validity .................................................................................................................. 24
3.7.4. Reliability ............................................................................................................................. 24

4. Results and Analysis ..................................................................................................... 25

4.1. Annual Reports ................................................................................................................................................... 25
4.2. Public News and Articles ................................................................................................................................. 25
4.2.1. AI Embedded Solutions for Customers ............................................................................... 25
4.2.2. AGROSAVIA Case ................................................................................................................. 26
4.2.3. Q&A with CIO of Thermo Fisher Scientific by CIO.com ....................................................... 27
4.3. Interviews............................................................................................................................................................. 28
4.4. Analyses of the Results ..................................................................................................................................... 36
4.4.1. Analysis of Secondary Data – Annual Reports, Public Articles and News ........................... 36
4.4.2. Analysis of Primary Data – Interviews ................................................................................ 37

5. Discussions .................................................................................................................... 45
5.1. Answers to Research Questions.................................................................................................................... 45
5.2. Limitations ........................................................................................................................................................... 47
5.3. Conclusion and Future Research ................................................................................................................... 47

References ............................................................................................................................ 49

Appendixes ........................................................................................................................... 57
Appendix 1: Interview Questions ................................................................................................................................ 57
Appendix 2: Interview Notes........................................................................................................................................ 58
Appendix 3: Diagram of 1st Order Concepts and 2nd Order Themes ................................................................ 86
Appendix 4: Diagram of 2nd Order Concepts and Aggregated Dimensions ..................................................... 87

iv
List of Tables
Table 1: Matrix of Impacts of AI Applications on R&D Management ................................................................. 12
Table 2: Matrix of Dimensions of Interview Questions ......................................................................................... 20
Table 3: Coding of Interview Data – 1st Order Concepts summary ................................................................. 35
Table 4: 2nd Order Themes summary ....................................................................................................................... 38

v
List of Figures
Figure 1: Process Improvement Model for Managerial R&D Activities .............................................................. 12
Figure 2: Thermo Fisher Scientific R&D Expenses in billion $ (2015-2020)...................................................... 17
Figure 3: Grounded Theory Steps .............................................................................................................................. 23
Figure 4: R&D Expenses vs Total Revenue in billion $ (2015-2020) ................................................................... 25
Figure 5: Illustration of Interactions in the AGROSAVIA project ....................................................................... 27
Figure 6: From 2nd Order Themes to Aggregated Dimensions ............................................................................ 39
Figure 7: Formulated Grounded Theory – Dynamic Model.................................................................................. 44

vi
1. Introduction

In this part we would like give readers a short but meaningful “time travel” ride from the birth of
computer to the current age of AI. As AI applications become increasingly pervasive in various sectors, it
is changing our work and life, especially in the business world. The impact “force” of AI applications on
management and leadership is shifting and reshaping our work environment and more importantly, our
mindset and behaviours in business. Therefore, to understand this impact “force” becomes a highly
interesting topic for business organizations and their leaders who want to be successful in this shift in
leading their organizations and people to bring more values to their customers quicker and eventually
become more competitive than their counterparts. We have already seen how AI applications are
successfully being used by organizations to enhance the value their products give to their customers.
Tesla is one outstanding example that has utilized AI and machine learning in autonomous driving of
their cars. This includes features like navigating not only on the freeways but also through local streets
as well as traffic signals (Rangaiah, 2020).
Besides the background, a rationale of the formulation of our research questions and definitions of the
related terminology will be presented in this section.

1.1. Background
During the interval between 17th and 19th century, the word “computer” was still having the same
meaning as human calculator in different work sections since there was nothing else in the planet doing
computation in an autonomous way. With Alan Turing’s groundbreaking work in modern computer
science in 1940s, the word “computer” became gradually detached from its original anthropological
meaning and instead turned to be a programmable machine with general purposes. The term “artificial
intelligence” was then first coined by John McCarthy in 1956 in an academic conference (Anyoha, 2017),
which has a myriad of definitions nowadays. One simple definition is that AI is the simulated human
intelligence by computers or machines which are programmed to rationalize as humans do and mimic
humans’ actions (Investopedia, 2021).

Humans are living in a world where technological advancement is progressing with a faster-than-ever
rate with an abundance of information and data being created and stored. It is estimated that the
worldwide data volume reached 59 zettabytes in 2020 and is projected to 149 zettabytes in 2024 (Host,
2021), where 1 zettabyte equals to staggeringly 1021 bytes. It is no doubt that we have entered an age of
big data where AI combined with big data opens up many possibilities that would be limited for the
human brain to process.

It has already been witnessed that AI is used in many different industries and applications to increase
the possibility of success for a firm. If used correctly, AI can be a very powerful tool helping to increase
the efficiency and success rate of many roles, including developing innovative management skills.
Leaders with innovative leadership capabilities are critical in a high paced and dynamic field of work. An
innovative leader has the skills to combine different leadership styles to influence employees to develop
creative ideas, services and products (David J. Hughes, 2018). Having creativity in an organization will
lead to innovations in services, processes and products. Therefore, all leadership practices that enhance
creative problem solving will enhance innovation.

1
A leader’s combination of attitudes and behaviors will lead to a specific predictability and regularity in
dealing with his/her group members. The definition of leadership style is the relatively consistent
pattern of behavior that characterized a leader (DuBrin, 2013). In literature, there are 3 leadership
theories that are being discussed the most: Bass (1985), Burns (1978) and Yukl (1989) describe
respectively transformational, transactional and laissez-faire leadership styles. Pearce et al. (2014)
present leadership as something that is shared and that it is just simply a matter of degree. This
leadership theory includes task/goal-oriented, relation/people-oriented and Laissez-faire (Kleefstra,
2019). Transformational leadership is highly related to innovation and a transformational leader is one
who brings out positive, major changes in an organization which is done by moving group members
beyond their self-interest and towards the best of the group or organization (DuBrin, 2013).
Transactional leadership influences their employees based on exchanging benefits, which can be both
internal and external motivators. There are two components that are included in transactional
leadership, management-by-extinction and contingent reward. Transactional leadership is often
contrasted against transformational leadership. Laissez-faire leaders are described as leaders who do
not take charge of their leadership. Laissez-faire leaders believe that people are uncontrollable and
unpredictable and therefore understanding people is a waste of time. Normally, this type of leaders
keeps a low profile and works with whatever structure that is available without any suggestions or
criticism (Kleefstra, 2019). The essence of transformational leaders is to develop and transform people
which will bring out the potentials and capabilities for innovation in the organization. The research
conducted by Hay Group (2005) for the World's Most Admired Companies, found that a major
characteristic for those companies is constant innovation. Innovative behavior of employees is
considered to be of significant importance for organizational effectiveness and for the survival of the
organization. It is important to achieve competitive advantage, especially in today’s rapidly changing
business environment and rapid technology advancement (Pieterse et al. 2010).

Among many functional departments of an organization, Research and Development (R&D) department
is the essential driving force for the long-term development of a firm. Innovation is a crucial component
for an R&D organization to improve technologies and products substantially to help the firm achieve
new competitive advantages (Keller, 1992). While employees think and implement innovative endeavors
in R&D, it is important to identify how management styles and behavior can affect the employees
related to their innovative abilities, especially with the presence of AI technologies.

1.2. Problem Formulation

The implementation of AI applications has become a world’s mega-trend in different industries
(Hufschmitt, 2019). Numerous success stories of AI implementations have just further accelerated the
expansion of AI applications (Joseph et al. 2015). Prominent examples are the reputable industry giants
such as Microsoft, Amazon, Facebook, Apple and IBM, which are among the most successful companies
in application of deep learning which is considered as a sub-branch of AI. Moreover, AI applications have
also been cascaded down to various functions of an organization, such as corporate program
management, finance, human resources (HR), information technology, legal and compliance,
manufacturing, marketing, operations, risk assessment, sales, strategy, supply chain, as well as R&D. A

2
number of studies show that significant successes are found in functions such as marketing (Sterne,
2017), manufacturing (Liu, 2020) and predictive maintenance in production operations (Cardoso &
Ferreira, 2021), as well as R&D especially with abundant examples in pharmaceutical industry
(Freedman, 2019). AI application in pharmaceutical sector has been successfully contributing to the
significant reduction of time in R&D for drug discovery, where a prominent example is that
pharmaceutical companies such as Celgene, GSK, Sanofi, and Sunovion partnering with Exscientia which
is specialized in AI Drug discovery, have successfully reduced the time needed for drug discovery
significantly from 4.5 years to just one year with cost reduction of 80% (Thermo Fisher Scientific, 2020).
But this does not mean that AI applications always embrace tailwind since challenges and hindrances
are inevitable. For example, human resources management is faced with tricky challenges to achieve
effectiveness by using AI due to HR management’s high level of innate complexity like measuring
employee performance, and the comparatively rare events such as recruitment and dismissals, for
which AI is not mature enough for bring effective and efficient solutions (Tambe et al. 2019).

At the same time, with the fast advancements with application of AI in the business world, it is believed
that a shift in leadership style is necessary in the short future since AI is believed to influence leadership
by taking over the “hard” elements of leadership such as profound expertise, decisiveness and
authoritative power, while the increasing emphases of leadership will be drawn onto the “soft” parts
like humility, adaptability, vision and permanent engagement (Chamorro-Premuzic et al. 2018). Different
studies have been done in looking into the possibilities to influence an organization’s or a leader’s ability
and decision-making when applying AI, while AI applications become pivotal in leaders’ pursuit of
success in innovation in intensely competitive environments (Shrestha et al. 2019; Metcalf et al. 2019).
The “question” is emphasized less on whether or not AI is impacting how firms and leaders make
decisions and interact with their stakeholders such as employees, suppliers and customers, but it is
clearly more about how AI should be applied to create the best values for the firms and the leaders.

Many factors have been found to affect organizational innovation, in which leaders and executive’s
leadership styles have been identified as a one of the most important factors. Despite this, we have
observed that few studies have been done in the linkage between innovation in the organization and
the leadership styles of the top managers.

Transformational leadership, in contrast to transactional leadership, has been argued to be particularly

effective in engendering followers’ innovative behaviour. Innovation is central to the thinking about
transformational leadership (Pieterse et al. 2010).

As above, understanding what changes AI technology applications have brought to or would potentially
bring to transformational leadership elements and the work of employees becomes a prerequisite step
to take on the path to achieve success in innovation. Therefore, it is of this study’s intention to dig
deeper to create some new insights in this topic which firms would need in order to possibly set new
strategies and reallocate their limited resources in the near future. Among different functions of an
organization, we are especially interested in the R&D function due to the following reasons:

3
- An organization needs to stay alert to innovative possibilities, as new disruptive technology can
make an industry obsolete or place a company in danger. R&D department is considered to be
strongly connected to innovation for a company to develop and introduce new products and
services to market. As a result, R&D is the “backbone” for innovation success.
- Enormous spending is usually strategically budgeted in R&D and its results can sometimes mean life
or death to a company. A report on pharmaceutical industry in 2019 showed that the global R&D
spending in pharmaceuticals was $165 billion in 2017 and is expected to increase by 3% per year,
staggeringly reaching around $204 billion by 2024 (Singh, 2019).
- AI has been adopted to significantly improve R&D efficiency, creativity and productivity. One
example is AI application in drug discovery process in pharmaceutical companies, including drug
repurposing, clinical trial designing and predicting of safety.
- There are very limited studies investigating the impacts of AI application leadership and
management in an R&D department, which is considered by us as a gap of key insights which are
needed for R&D leaders to have in order to plan and allocate resources, as well as to set up and
deploy effective strategies to achieve corporate success in innovation. Besides, there is little
research conducted in the field of R&D related to leadership and innovation which is important for
success in R&D work because leadership behaviour is used to motivate and direct professional
employees.
In order to explore this uncharted area and create new understanding and insights, we decided to select
a target company (Thermo Fisher Scientific) to conduct a case study. Hence, the research first question
(RQ) of our study is:

RQ1: How do AI applications impact on management and leadership in R&D department in Thermo
Fisher Scientific?

Having attempted in answering the first research question, we have taken one step forward based on
the understanding on the impacts of AI on R&D management and leadership, to investigate what
concrete suggestions will be applicable for leaders to take for innovation success while implementing AI
in R&D of the target company. This leads to our second research question:

RQ2: How can AI applications be utilized to stimulate innovation in R&D department in Thermo Fisher
Scientific?

1.3. Definitions

1.3.1. Artificial Intelligence in a Nutshell

There exist numerous definitions of AI and many of them can be either incomplete or elusive. The
simple definition shared in Section 1.1 that “AI is the simulated human intelligence by computers or
machines which are programmed to rationalize as humans do and mimic human’s actions”
(Investopedia, 2021) can also be somehow controversial. Due to this ambiguity, it is crucial for business
leaders to first understand what AI can do in application, including its capabilities and limitations. In
terms of capabilities, artificial intelligence can be categorized as 3 different types from low to high
capabilities as follow: Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI) and
Artificial Super Intelligence (ASI) (Miailhe & Hodes, 2017). However, in terms of universality of utilization,

4
ASI is still far away from reality and is hypothetically fictional, while AGI has just achieved some
theoretical foundations in recent years. However, ANI has been the most widely used and has achieved
tremendous successes in different business sectors.

1.3.1.1. ASI, AGI and their unresolved problems

ASI is regarded as the highest artificial intelligence which is superior to human intelligence and cognitive
ability. In contrast, AGI is known as artificial general intelligence, which is also known as strong AI,
referring to intelligence equivalent to human-level intelligence (Miailhe & Hodes, 2017). This means
whatever human can do a strong AI is capable of doing that as well. However, until recently both ASI and
AGI are still not achievable in the real world. There are two major problems which are hindering the
success in ASI and AGI, where one is called the frame problem and the other is symbol grounding
problem (Floridi, 2014). The frame problem refers to how a machine can represent the changing
surrounding and have correct interactions with it by time, while the symbol grounding problem refers to
how data acquire their semantic meaning. So far there has been no clear clue on how the frame
problem and the symbol grounding problem can be solved without human interventions and
manipulations.

Nevertheless, there has been optimism around that the journey to AGI has embarked with the recent
development and presence of the algorithms for answering casual and counterfactual queries, and the
only thing left is the implementation of those algorithms by industrious AI researchers (Pearl &
Mackenzie, 2018).

1.3.1.2. ANI
With the 2 major hurdles aforementioned to be overcome in the future, even the most intelligent
seeming AI applications in realistic use today are only ANI, which is also known as weak AI (Miailhe &
Hodes, 2017). There are two big categories in weak AI: symbolic learning and machine learning.

Symbolic learning

Symbolic learning dates back to its birth almost six decades ago when the field of AI was dominated by it,
also known as “classic AI” or “rule-based AI”. Symbolic learning is one way for computers to process
information through the articulated embedding of human knowledge and rules into computer programs
(Dickson, 2019).

Symbolic learning is highly convenient for settings in the circumstances in which rules are crystal clear
and symbols can be transformed easily after obtaining inputs. The natural language processing tool Eliza
and the General Problem Solver program were the famous successful stories between 1950s and 1960s
(Haenlein & Kaplan, 2019). And nowadays it is still widely used in robotics.

However, symbolic learning is very fragile because it is extremely restricted by the rules. Any changes in
the working assumptions will demand that the corresponding rules be rewritten. The rewriting process
is enormously laborious, and it is simply infeasible to define rules for everything in the world with

5
constant changes. In this sense, future development of rule-based AI is argued to be the wrong track
(Pearl & Mackenzie, 2018).

Machine learning

Machine learning is another category of AI which is aimed for data-based pattern recognition by using
abstruse statistical techniques enabling computers to improve performance by experience. Computers
are efficient in pattern recognition due to its strong capability in utilizing more data and different
dimensions of data. Machine learning can be subdivided into statistical learning and deep learning.

Statistical learning refers to statistically based pattern recognition from data which is widely used in
speech recognition and Natural Language Processing (NLP).

Deep learning refers to the way to recognize pattern from data by trying to replicate the structure and
functionalities of the human brain to hopefully achieve the cognitive capabilities in computers. As a
result, deep learning is a form of artificial neural networks. In its application deep learning is highly
complex and deep by using layers of neural networks to understand complex things. One famous and
prominent example of the high performance of deep learning is that AlphaGo, which is a program
developed by Google, defeated the world champion of the traditional board game “Go” in 2015. Go was
believed to be a board game in which computers would impossibly defeat human due to its high
complexity and abundance of possibilities, for example at the opening of the game there can be 361
possible moves in GO whereas there are only 20 for chess. One year later in 2016, AlphaGo’s upgraded
version AlphaGo Master achieved a staggering row of 60 full wins out of 60 matches against professional
Go players. Until today, artificial neural networks and deep Learning have established the foundation of
the biggest majority of applications we get to know in AI.

1.3.2. Transformational leadership

Leadership dates back to our early ancestors and have existed for a very long time. However, the shape
and skills of leadership have been reshaped during its history and this is believed to further develop in
the coming future (Thaddeus Hunt, 2018). Leadership is considered to be the most important aspects of
management and both huge international organizations like GE, Amazon but also great nations like the
USA, Britain and India are some examples of prominent nations that have emerged with the help of
effective leadership. Leadership have been defined in many different variations; two definitions are that
it is “the accomplishment of a goal through the direction of human assistants” (Prentice, 2004) and “the
process of social influence, which maximizes the efforts of others, towards the achievement of a goal”
(Kruse, 2013).

Transformational leadership is a theory where the leader stimulates and inspires their followers to
achieve extraordinary outcomes. This is done by paying attention to the developments needs and
concerns of the individual followers. Transformational leadership was first introduced by James
Macgregor Brins in 1978 (Odumeru & Ogbonna, 2013). Another definition of transformational
leadership is “a style of leadership that transforms followers to rise above their self-interests by altering
their morale, ideals, interests, and values, motivating them to perform better than initially expected

6
(Pieterse et al. 2010). Transformational leaders are able to align the followers' desires and needs to the
organization interests and goals.

1.3.2.1. Transformational Leadership and Innovation in R&D function

Innovative behaviour is defined as “a multi-stage process to recognize problems, generate ideas and/or
solutions, building support for ideas, and idea implementation”. Creativity that is defined as “the
production of novel and useful ideas” is an important part of innovative behaviour (Pieterse et al. 2010).
It is believed that transformational leaders stimulate followers’ innovative behaviour by expressing an
inspiring vision, stimulating the employees to think outside the box and question the status quo. It is
important that the followers go the extra mile to be able to reach innovative behaviour, and this is
believed to only possible when you align the interest of the organization with the needs and desire of
the followers. Only motivating the employees with transactional methods and contingent reward (clear
relationship between expectations and reward) will most probably not promote them to go the extra
mile, which is needed to reach innovative behaviour.

Transformational leaders, in contrast to transactional leaders, put more emphasis on innovation than
transactional leaders and research have found a positive relationship between organizational innovation
and translational leadership (Odumeru & Ogbonna, 2013; Pieterse, et al., 2010)

R&D organization is the innovation powerhouse for an enterprise to develop sustainably and innovate
continuously to maintain its core competitiveness. R&D is therefore considered as the essential strategy
for a firm to seek competitive advantage. The main functions of R&D include development of new
products/services by adopting new technologies or novel knowledge, and improvement or
enhancement of existing products/services to meet changing market needs in both short and long terms.
A firm’s senior leadership team in R&D plays a critical role in forming the firm’s competence to exploit
the existing capabilities and explore the uncharted technological territories (Tushman, 2017). The result
of a large-scale survey shows that compared to non-R&D managers, R&D managers tend to be more
creative and quicker in mastering technical knowledge, but generally are worse in managing individuals
in situations where dealing with conflicts and creating a creative and welcoming work environment are
needed in transformational leadership (Gritzo et al. 2017). The relationship of more innovation for
transformational leaders doesn’t necessarily mean that transformational leaders promote innovation. It
could also be a result of strategic decision of for example investing more in R&D. Studies have shown to
increase performance quality in R&D relates to transactional leadership and followers’ innovative
behavior (Keller, 1992). This can be argued to be mostly dependent on innovative behavior.

1.4. Delimitations
AI, leadership and innovation are all very broad research topics and the combination scope of these
focus areas that can be chosen is even broader. This is a reason why we limited our study as a first step
to find a focus area that we believe is of most interest and that would have the biggest impact on
organizations in the future. We identified Artificial Narrow Intelligence (ANI) which is currently the most
widely and successfully used AI as the target type of AI in application, disregarding the Artificial General
Intelligence (AGI) and Artificial Super Intelligence (ASI) which are far from mature and commercialized.

7
So, when we mention “AI” throughout the report in later chapters, we refer to ANI. In addition, we
limited the study to leadership and management in R&D setting in the target company Thermo Fisher
Scientific as we consider it is most relevant study object regarding innovation.
Data from different sources collected in our qualitative and exploratory study will include subjective
perceptions, for examples, from perceptions of interviewees. All the collected data will be limited in our
target company Thermo Fisher Scientific in life science sector, and it is therefore limited for
generalization.

1.5. Structure
The thesis consists of the following main chapters:

Chapter 1, Introduction – This section gives a brief introduction of the history of AI development and the
phenomenon of its increasing popular use in different industries as well as its potential influence on
leadership and innovation especially in R&D. Research questions are formulated based on the
phenomenon as the research purpose of this study.

Chapter 2, Literature Review – In this section, we present the theoretical background to the problem
area, as well as an introduction of our theoretical framework for the study.

Chapter 3: Methodology - In this section, we present the choice of research method for the study and
our research design. We also motivate here the reasons for the choice of method as well as how data
was collected and later processed.

Chapter 4: Results and Analysis– Here we present our findings of the empirical study from our data
collection from different sources. We support our findings with tables and figures where appropriate.
We dissect our findings into analyses in a structured, systematic and logical manner by using grounded
theory approach. Eventually a dynamic model is created and presented based on the primary data from
the interviews.

Chapter 5: Discussion – In this section we discuss how our research questions are answered based on
the data analyses and the reviewed literature, and further discuss the limitations that are inherent in
our study, as well as the conclusion of implications to future interested researchers and practitioners.

8
2. Literature Review

This part of the study is a literature review, based on which our theoretical foundation on the two
research questions will be established and formulated. A theoretical framework is then established
based on a thorough review of existing theories and empirical studies, which will be used as our
operationalization model in this study.

2.1. The Impacts of AI Application on Leadership and Employees in Business

Intelligence, or cognitive ability, has traditionally been considered as an important element for
successful leadership. There is seldom a lack of studies supporting this viewpoint. In fact, the
relationship between intelligence and leadership can be regarded by some researchers as “common
sense” (Fiedler & Garcia, 1987). In one comprehensive review, intelligence is proved to be one of the
best predictors of general job performance and it is argued that intelligence is outperforming any other
leadership traits in predicting numerous influential real-world outcomes, even the trait of
conscientiousness (Schmidt & Hunter, 1998). With big data and AI gaining foothold in the business world,
decisions of leaders based only on intuitive and previous expertise are becoming less and less credited
(Economist Intelligence Unit, 2012). But we need to be aware that although AI applications are widely
used in almost all functions in businesses, they are still ANI and there is no reason to believe that human
leadership will be spared in a foreseeable future because many traits of leadership cannot be replicated
by AI yet. However, the external context changes are undeniably to some extent reshaping leadership
behaviors and skills which are critical for leadership effectiveness (Norris & Vecchio, 1992). In a research
co-initiated by Institute of Management Development (known as IMD) and Cisco, it concludes that the
winners in the current drastically changing “digital vortex” environment will be the organizations which
have sufficient agility to innovate promptly and unleash their capability to produce cost value,
experience or platform value for their customers (Joseph et al. 2015). Another recent research suggests
that the essence of effective leadership should be rethought since some “hard” elements of leadership
such as profound specialty expertise, decisiveness, authority and short-term focus on task would assume
to be gradually taken over by AI, while the “soft” elements such as humility, adaptability, vision and
engagement will shift to be the key elements in more-agile leadership (Chamorro-Premuzic et al. 2018).
Another study also echoes on the unprecedented trend on increasing focus on people-focused tasks in
the workplace, due to the fact that AI applications are doing very well in mechanical intelligence and
thinking intelligence, but not in the feeling intelligence which requires the capability in recognizing,
emulating and responding to human emotions in an appropriate way (Huang et al. 2019). This is due to
the AI’s excellent efficient capabilities in data processing and analysis and decision making (Shrestha et
al. 2019). However, for the non-cognitive parts, AI is still lagging behind. Meta-analyses studying 50
years of research indicate that personality traits, such as emotional stability, extraversion, openness,
agreeableness and integrity, are twice as importance as intelligence in predicting leadership
effectiveness (Judge et al. 2002, 2004). As a result, leaders probably need to shift the emphasis of their
team from cognitive ability to emotional, empathetic and interpersonal relationship skills.

9
2.2. How AI Can be Best Utilized by Leaders to Achieve Innovation Success in R&D
Based on the first research question, as AI applications are expected to influence leaders and employees
in R&D, the next practical question will naturally be how AI can be utilized to bring the best innovation
values in R&D function to leaders and the firm. This question is highly relevant for business practitioners
who are planning or have initiated AI utilization into part of their strategies, no matter if it is for
achieving higher operational efficiency, higher revenues, better customer satisfaction and stronger
competitiveness. In a recent valuable study carried out by Brock&VonW. (2019), the authors conducted
a large-scale survey on 538 firms and compared the successful and unsuccessful firms in implementing
AI-related projects to see what makes the performance differences. Based on their analyses and
insights, they have identified seven guidelines for managers and leaders to act and implement when
trying to get success in AI applications, which are condensed into an acronym “D-I-G-I-T-A-L” standing
for being focused on Data, being Intelligent, being Grounded, being Integral, being Teaming, being Agile
and being focused on Leadership (Brock & von Wangenheim, 2019). As for R&D domain, the guidelines
and ethical risks in AI applications should be kept in leaders’ mind, with a higher resolution and
granularity from R&D perspective in order to achieve success in innovation. It is undeniable that AI
ethics is highly relevant for companies of different industries, especially those with advanced AI
implementation, and governments, as well as international institutes and organizations to work
together in order to find the best practical ways to make AI development and application compliant to
our ethical values. A recent study by Hagendorff (2020) examines and analyses 21 of the existing major
AI ethics guidelines and concludes that they are with very low effectiveness when deviations of the
ethical codes of conduct occur. This is echoed by a recent controlled study that the effectiveness of
codes or guidelines of ethics is almost non-existent, and they simply do not change at all the behaviour
of the sampled software engineers from the tech community (McNamara, et al. 2018). Schuklenk (2020)
argues that government funders or non-partisan charitable organizations could possibly provide a
counterbalance against the generous tech giants who have substantially funded academic competence
and credibility to support their AI ethical policy development. As for our study, our main focus is on
exploring how AI application impacts R&D management and leadership, as well as innovation by a case
study of our target company. It is admittedly clear that ethical consideration would affect leaders’
decision making. However, AI ethics and its definitions and boaderlines are under development and
dependent on large scale of international efforts of differnet groups and organizations. As above, AI
ethics will not be a focus of our exploratory case study of a company not yet starting with AI
implementation and we assume that the AI development and application should be accoridng to the
existing ethical guidelines if AI application is implemented.

Based on the above literature review, we have decided that a case study will be the most appropriate
method of choice. Due to the nature of our research question, too many variables will be under study,
and therefore conduction of a case study would be the best approach.

2.3. Theoretical Framework

Our case study is an exploratory research and there is currently limited understanding in academics and
empirical studies. As a result, multiple sources are expected to be used to collect huge amount of data,

10
such as primary data from empirical manifestations, secondary data like annual reports and company
news, to understand the phenomenon. However, through the huge amount of data obtained, we need
to have a theoretical framework in place so that we will have a “knowledge-base” and a “strategy” with
which we can interpret the correct messages and map the “reality” from the massive amount of data. In
order to create proper explanations or theories, we need to relate them to relevant knowledge which is
based on the use of concepts and theory (Ghauri et al. 2020).

Among different organizations, “people, process and technology (PPT)” have been commonly used as
the key dimensions in change management for a long time (Maltaverne, 2015). In other words, PPT’s
major contributions are for process improvement in management. Despite its universal use, PPT’s origin
is unclear. One possible recognized trace was Leavitt’s Diamond Model in 1964 which is including four
components: structure, technology, task and people (Cooper et al. 1964). One of the first appearances
of application of this model was within information technology (IT) industry in 1980s where the
framework of Information Technology Infrastructure Library (ITIL) was launched, which was including a
series of detailed practices concentrating on the alignment of IT services with the business value
delivery (Office of Goverment Commerce, 2007). In our study, PPT model is highly relevant as the three
key different dimensions in the knowledge-and-innovation intensive R&D department. However, PPT
model has been challenged by some other models for improvement purpose. Pearson Model is one
example by including a new dimension of information which is considered to be independent from
technology (Pearson, 2009). Additional dimensions have also been argued as valuable such as
organizational culture and strategy (Buttles-Valdez et al. 2008; Pisano, 2012). Prodan et al. (2015) have
argued that PPT model with additional three dimensions which are “customer focus, management
functions and innovation” will strengthen business value delivery by an intensified focus on customer
demands and innovation (Prodan et al. 2015). In our study in R&D department, apart from the three
aforementioned classic key dimensions of people, process and technology, we will include “strategy” as
the additional dimension we aim to study because:

1. The multi-facet nature of strategy indicates that strategy is about “courses of actions and
resource allocation to carry out to achieve long-term goals and objectives of the business
enterprise” (Chandler, 1962), “a different set of activities to deliver a special mix of value”
(Porter, 1996) and “the result of executive decisions made on where to compete and how to
compete to win, to maximize long-term value” (Favaro et al. 2012). All in all, it is clear that
strategy is essential and fundamental to an organization’s prosperity (Besanko et al. 2016).
2. Innovation performance is hinged on superior R&D performance which is dependent on a good
R&D strategy which provides consistency, coherence and alignment of a series of decisions
(Pisano, 2012).

Therefore, with the effect of “strategy” in harmonization with the PPT model, a new model is created for
managerial R&D activities visualized as below in Figure 1:

11
 Figure 1: Process Improvement Model for Managerial R&D Activities

Big data and digitalization are considered to have three types of influences on management, which are
informing, enabling and transforming (Blackburn et al. 2017; Farrington & Alizadeh, 2017). There is a
strong connection between “informing” and “enabling” so we consider “informing and enabling” as a
combined dimension for analysis. These three influences are highly compatible and applicable when it
comes to interactions between AI applications and management, therefore they will be considered and
used in our construction of the theoretical framework.

Based on the above extensive research review, a clear picture of the theoretical framework is created as
the below matrix (Table 1) for mapping the impacts of AI applications on R&D management:

Management (R&D)

AI Impacts /
Management People Process Technology Strategy
Dimensions

How will AI applications

inform and enable new
R&D activities?

How will AI applications

transform traditional
R&D?
Table 1: Matrix of Impacts of AI Applications on R&D Management

12
This above framework will be used as the guidelines in planning and executing the data collection
process and the subsequent data analysis. Based on the analysis results, interpretation will be
conducted to understand the relevant impacts on leadership, which was conceptualized as four
components including idealized influence, inspirational motivation, intellectual stimulation and
individualized consideration (Bass, 1985). According to Bass, idealized influence is defined as charisma
and behavior of role modeling, while inspirational motivation refers to providing meanings to followers’
job through visualizing convincing and appealing vision for the team. Intellectual stimulation is defined
as constant encouragement to followers to challenge current existing assumptions and ways of working
by rethinking the problems and creating new ideas, while individualized consideration refers to the
behavior of coaching and mentoring toward team members.

13
3. Methodology

In this part, an introduction of our selected method – case study, will be presented with justifications.
Moreover, our research design, data collection, as well as the corresponding validity and reliability will
be thoroughly discussed.

3.1. Selection of Research Approach

Case study is often associated with exploratory and descriptive research. However, it is not necessarily
limited to these areas. A case study is particularly useful when the phenomenon under investigation is
difficult to study outside its natural setting. It is also particularly useful when the concept and variables
under study are difficult to quantify (Ghauri et al. 2020). A case study is not suitable for all types of
research questions and it is the objective of the research and the RQ that decides whether a case study
is the appropriate method. A case study is useful for theory development and testing the theory (Ghauri
et al. 2020). It is appropriate to use a case study when trying to answer the “how” and “why” questions,
when the researcher have little control over events. The nature of a case study is to be exploratory,
explanatory or/and descriptive. It is appropriate to use this method when you are entering a new
research area or when the existing theory for the areas of research is inadequate. A comprehensive case
study can for example involve a study of several organizations in regard of a set of variables that already
have been identified. This type of study can be done to compare a phenomenon over different
organizations to be able to distinguish between them (Ghauri et al. 2020). In our case, in regard of our
research questions, we are more interested to explore a certain field, therefore we have not chosen a
comprehensive case study. Yin (2009) suggests that a case study is the preferred method when you
want to study some specific characteristics of a rare or extreme situation in which an organization finds
itself or if we want to study a situation within a firm that has rarely been studied and is unique in its
nature. Our research questions are considered to be relatively unexplored and rare, therefore we have
chosen a research method that is explanatory, exploratory and descriptive in its nature.

The strengths and weaknesses with case study as a research method:

Adelman et al. (1976) suggested that there are six main advantages with using a case study as the
research method:

1. This research method is down to earth and can often appeal to practitioners that will be able to
identify the issues and concerns raised, due to that case studies as a research method is strong
in reality.
2. Generalizations can be made from case studies.
3. Can offer support to alternative interpretations as it represents a multiplicity of viewpoints.
4. Provides a data base of materials that can be interpreted by future researchers.
5. Results and conclusions obtained from a case study can be immediately useful for a variety of
purposes.
6. Case studies are normally more accessible than other conventional research reports, making
them more capable of serving multiple audiences.

14
Merriam (1998) proposed three additional benefits by using a case study.

1. Case study is the best method for answering research questions and its strengths outweigh its
limitations. This research methodology offers a means of investigating complex social units that
is consisting of multiple variables of interest for the ability to understand the importance of the
phenomena.
2. Case study is particularly a choice of method of applied fields of study such as education.
Processes and problems can be examined to be able to understand how it in its turn can affect
and improve the different practice.
3. It is useful for studying educational innovations, for evaluating programs, and for informing
policy.

However, like all other research methodologies, case study comes with both strengths and weaknesses.
Guba & Lincoln (1981) presented in their study the limitations of having a case study as a research
methodology.

1. A case study can lead the reader to erroneous conclusions about the actual state of affair by
either oversimplifying or exaggerating the situation.
2. A qualitative study (including a case study) is limited by the sensitivity and integrity of the
investigator as the researcher is the primary instrument for the data collection and data analysis.
3. The case study evaluation can be related to the problem of ethics. A researcher that is unethical
could choose from available data and be selective in the content to present. Therefore, both the
reader of the case study and the researcher needs to be aware about this risk of bias and how it
could affect the final product.

Like with some of the other research methodologies, validity and generalizability is also something that
needs to be taken into account. Case study have been faulted for its lack of representativeness.
Collection, construction and analysis of the empirical materials are parts that the case study can lack
rigor in.

3.2. Research Design

The nature of a case study is to review existing historical material and records and combine these with
interviews to explore further in the matter of interest. Therefore, we intend to choose a public company
to study and perform the interview with leaders working in an R&D setting.

In contrast to performing surveys, where data collection is routinized and there is no need for the data
collector to have certain skills, for a case study, it is important that the person collecting the data have
certain skills in the dynamic of the case so that they can grab the opportunities presented to them
during the data collection process (Ghauri et al. 2020).

Bonomas’s theory/data/theory revision process:

Bonoma (1985) suggested that it is useful to perform a theory/data/theory revision cycle when
performing a case study. This process includes 4 stages, drift, design, prediction and disconfirmation.

15
We have therefore included this process model for our case study. The drift stage is done in the
beginning of the research study and consists of learning the research area including the different
concepts and terminology that is being used in the field. The aim of this stage is to widen the
perspective of the researcher, which often leads to modifications in the original research question. The
next step in this revision process is the design phase, where the researcher chooses the strategy to
collect the data that is needed to be able to answer the research question. In this stage, the researcher
also starts to develop provisory explanations for the observations that have been collected. It is in this
stage that the researcher assesses and refines the major areas of the research project the originally was
suggested in the drift stage. This is a dynamic process where the researcher can go back to the drift
stage for better understanding. The middle stage of the revision process includes the prediction stage.
By the time the researcher has reached this stage, he/she should have a good understanding of the
factors that the case information may be grouped within and can therefore proceed to the construction
and analysis part. In this stage, the researcher should also confidently proceed with compiling more
cases to be able to draw conclusions. Therefore, this stage might also include the development of
tentative explanations. Disconfirmation that is the last stage, include further testing suggestions and
analysis of the result that was suggested in the prediction stage. The researcher can do this by applying
the result to other cases or to a broader set of cases. The purpose of those four stages is not to form
some kind of hierarchy, but rather to form an iterative evolution towards getting an understanding
(Ghauri et al. 2020, pp. 105-106).

3.3. Choosing Target Company for Case Study

3.3.1. Thermo Fisher Scientific as a Company

Thermo Fisher Scientific is the world leader in service science by providing scientific instrumentation,
consumables and reagents, software and services to life science, healthcare and other laboratories in
academia, government and industry (including biotechnology and pharmaceutical sectors). The mission
statement of Thermo Fisher Scientific is to make the world healthier, cleaner and safer (Anon, 2021).
Like many other firms in this field, Thermo Fisher Scientific is working in a dynamic field where
innovation has a main impact on the company's financial performance and competitiveness (Thermo
Fisher Scientific, 2021). According to Thermo Fisher Scientific’s 2020 Annual Report, there are 3 pillars
that support the company’s growth strategy (Thermo Fisher Scientific, 2021):

1. Continuation of developing high-impact and innovative new products

2. Leverage of scale in high-growth and emerging markets
3. Delivery of a unique value proposition to customers

16
3.3.2. R&D in Thermo Fisher Scientific
Thermo Fisher Scientific presents strong resolution and ambition in innovation which is considered as
the first pillar for its growth strategy. Over the past six years, the company has continuously been
increasing the R&D investment (Figure 2). During year 2020, R&D expenses of the company has reached
record high $1.2 billion in order for delivering new technologies to assist customer in achieving their
targets (Thermo Fisher Scientific, 2021).

R&D Expenses (billion $) increases over time.

1.4
R&D Expenses (billion $)

1.2 1.2
1.0 1.0 1.0
0.9
0.8
0.8
0.7
0.6
0.4
0.2
0.0
2015 2016 2017 2018 2019 2020
Year

Figure 2: Thermo Fisher Scientific R&D Expenses in billion $ (2015-2020) (Thermo Fisher Scientific, 2021)

COVID-19 pandemic has been the mega-background for the whole world in year 2020 and posing a
grave challenge for global communities, governments and businesses to individuals. In contrast, as a
provisioner of scientific instrumentation, reagents, consumables, as well as software and services for
different industries, Thermo Fisher Scientific has drastically increased its growth and created enormous
value for its stakeholders during the pandemic, with revenue exceptionally growing by 26% to $32.22
billion (Thermo Fisher Scientific, 2021). This increase is in strong contrast with the steady growth in the
previous years.

3.3.3. AI Development in Thermo Fisher Scientific

AI is having a big impact on the life science sector. Implementation of AI in this sector includes, drug
discovery, imaging diagnostics, drug development and precision medicine and more. For those reasons,
we have chosen to target Thermo Fisher Scientific as this firm is one of the big firms operating in the life
science field. Organizations operating in the life science sector would potentially benefit from AI due to
the nature of their products (Thomas Marshall, 2017). Thermo Fisher Software applications can support
the use of advanced digital technologies, such as AI, to accelerate science (Anon, 2019). Genome
sequencing together with relationship mapping between genes, clinical diagnosis/outcomes and
phenotype data are other fields that Thermo Fisher Scientific is operating in where AI would be highly

17
beneficial. Due to the size of the human genome of over 3 billion nucleotides, rule-based systems
(software) are not possible to use for analysing and provide insights about human genome applications.
However, AI and machine learning would be the appropriate tool to utilize, and there are already
examples of AI and machine learning contributing to disease diagnosis/prediction and drug
design/discovery (Shrestha, 2020). We did not find AI being implemented in different organization
across Thermo Fisher Scientific including the R&D department.

Our chosen case is the R&D department at Thermo Fisher Scientific, as this would be relevant for AI,
innovation and leadership. We will perform multiple interviews within the company’s R&D department.
Yin (1994) presented 4 different types of case study designs and presents a 2x2 matrix with 4 different
combinations. The primary distinction is between single and multiple case designs. A single case study
design can be used when performing a pilot study or an exploratory study. Single case study design is
also a preferred choice when the singe case is extreme or unique. This means that we can study and
observe a phenomenon that previously was not accessible. The choice of case study method also
depends on if we aim for an inductive or deductive approach, as we are aiming for an inductive
approach, a single case study is more suitable (Yin, 1994).

The case is chosen to correspond with our theoretical framework and the variables we are studying. As
we are studying the behavior of R&D and innovation, we selected a firm that is dealing with innovation
and R&D and selected leaders to interview who are involved in this process. With case studies, it is
important to interview the "right person" from the organization (Ghauri et al. 2020). The right person
means from the point of view of our research question and study variables. For a case study, often 1
case is sufficient (Yin, 1994).

3.4. Data Collection

3.4.1. Sources of Data

Multiple sources of data will be used in our study including primary data from multiple interviews to
relevant people working in R&D department and other departments in the target company Thermo
Fisher Scientific, and secondary data from existing historical material and records such as the subject’s
annual reports, relevant public news and other possible sources which are published openly.

Purposeful sampling of interviewees will be conducted in our study for primary data collection because
this approach is argued as most suitable due to the fact that the purpose of our in-depth case study is to
explore new insights and understanding in our research question instead of focusing on generalizability
(Patton, 2014). Purposeful sampling involves the identification and selection of knowledgeable and
experienced individuals or groups of individuals for a specific phenomenon (Cresswell & Plano Clark,
2011). However, Palinkas et al. (2015) argue that compared to probabilistic or random sampling,
purposeful sampling has inferiority in generalizability due to limited constraints for potential of bias in
selection and limited control for the potential influence of confounders which are known or unknown.
For our study which is aimed to explore and create new insights, purposeful sampling will suffice to
meet the purpose.

18
Semi-structured approach was used in the interview planning and executed in the interviews according
to our theoretical framework. This approach is more valuable in our exploratory study compared to
structured interview approach because semi-structured interviews provide us with possibility to get a
clearer and more accurate picture of real-life phenomenon from respondents’ answers, while the
strategy of set-up and sequences of the questions in semi-structured approach are prepared in detail in
advance compared to unstructured interviews without such preparations. However, semi-structured
approach will require that interviewers have more know-how and cautions during the conduction of
interviews and have complete understanding of the research questions. In order to fulfill this demand, a
comprehensive and structured interview guide will be created based on the theoretical framework to
make sure all questions and potential sub-questions will be relevant to our research questions. We
decided to apply an exploratory research due to the outcome of our theoretical framework. Our
understanding of this research question is that the problem is only partly understood. Therefore, the
goal of our study is by conducting interviews, gain understanding of how this question is understood in
the organization and or companies alike who have not yet started or planned to start with AI
implementation. All the interview questions are included in Appendix 1.

3.4.2. Sample Selection

As a target population, we have chosen Thermo Fisher Scientific as an organization. As a group and
element, we will focus on the company’s R&D department and leaders in this domain. The accessible
population will therefore be leaders in this firm working in the domain of R&D. To be able to work
within the scope of this study and the time available we aimed to select 3-4 people in one organization.
We planned therefore to conduct interviews on at least 3-4 leaders working within R&D. In big
organizations, it is important to select the right department, but the choice of section and the individual
is also important (Ghauri et al. 2020).

3.4.3. Basic Information of the Interviewees

For the interviews, we have chosen employees at Thermo Fisher Scientific from different departments
but with an emphasis on the R&D department due to the background of the study. We chose to include
employees from different departments to be able to increase the possibility to have a comprehensive
view and a constructive result of the study.

In average we have had an acceptance rate of around 23%. 9 of 39 asked persons agreed to participate
and conduct the interview.

In total there were 9 interviewees from different departments in the target company Thermo Fisher
Scientific participating in the interviews for data collection. One of the interviews was conducted with 2
persons at the same time, and we treated the data as two separate interviews even if we are aware that
the answers could be biased. 5 interviewees (3 of 5 are in managerial position) are directly working in an
R&D department while the remaining 4 are working as managers in different departments across the
organisation. The other departments included were Service & Support, Customer care and Marketing. 4
of the 9 interviewees have been working for the company for more than 5 years.

19
3.4.4. Interview Design
Considering the cost, safety and practicality during COVID19 pandemic, all of the interviews were
conducted though the video communications tool Zoom. Interview time frame is set to be 60 minutes
which is agreed with the participants in the confirmation emails and all interviews were finished within
the pre-set time frame.

We have initiated the interviews by having an introduction to both introduce ourselves to the
interviewee and get an introduction to the participants. Following that, we presented the aim and
intention of the study and what we are trying to achieve with both the interview and the study. We
continue with presenting the ethical considerations and most importantly ask the permission for
recording the interview. The recording of the interviews was for the purpose to process the interview
data and transcribe the result later.

We have performed pilot interviews prior to the interviews to ensure the interview questions would be
easy to understand and that the timeframe of conducting the interview was within our scope. The pilot
interviews were performed using the early draft of the interview guide and the outcome was used to
design the final interview guide that we used for the interviews. The data collection and data analysis
were done simultaneously as this being characteristic of grounded theory. Therefore, the interview
design was slightly modified during the process of data collection to better fit with our case study. After
conducting the pilot interviews, minor adjustments were done to the interview guide. The number of
questions included, and the order of the questions that was asked were affected.

The interview guide and the questions were divided into the following sections and asked in this order:

- Questions related to the management style and skills.

- Questions related to the different dimensional combinations (Table 2):

People Process Technology Strategy

Inform&Enable Dimension 1 Dimension 3 Dimension 4 Dimension 7
Transform Dimension 2 Dimension 4 Dimension 5 Dimension 8
Table 2: Matrix of Dimensions of Interview Questions

The theoretical framework was used to design the questions aiming to extract data that is in alignment
with the different dimensions stated above, together with the collection of management styles.

The questions were designed in an open-ended way with the aim to collect as open data as possible
with the minimal impact of bias in the answers and to retrieve plentiful data for the different
dimensions. Because of this, the questions were considered to be difficult according to feedback from
the pilot-interviews. The target of the interviews is to collect as much relevant information as possible
by asking the listed 30 semi-structured questions. However, some questions were selectively unasked
for people who do not have actual working experience in R&D, such as questions regarding AI impacts
on R&D processes. The notes of the interviews are included in Appendixes part (Appendix 2), with the
unasked questions marked.

20
3.5. Data Analysis
We have used the grounded theory approach in our research study. Grounded theory involves the
construction of theories and hypothesis from analysing collected data. It is described as applying
systematic and analytical approach through the study of reality without any pre-assumed understanding
(Gioia, et al. 2012). We therefore start from the data and let inspection of the data result in explanation.
By using grounded theory, concepts are being created instead of developing constructs. In comparison
with constructs, concepts are more general and less specified. Therefore, they can describe and explain
a phenomenon better (Gioia & Pitre, 1990).

In grounded theory, concepts are considered better to develop before developing construct for the sake
of theory development. Interpretations are done by in a systematic way presenting a first-order coding
(analysis) that is based on the informants’ understanding and terms. A second order analysis that is
based on the researcher’s concepts and themes is generated. The second order analysis is therefore
when the researcher does the interpretation and get the concepts into themes. The third and final step
is then to further analyse the themes to aggregate dimensions (Strauss & Corbin, 1998). These are also
called the sequential series of stages that is divided into three stages: open coding, axial coding and
selective coding. The open coding is the procedure for developing the categories of information while
the axial coding is the procedure for interconnecting the categories. The final stage is the selective
coding that is the procedure for building a story that connects the categories and that will produce a
discursive set of theoretical propositions (Strauss & Corbin, 1998). There are also other processes that
can be followed, and the researchers are not limited to the stages presented by Strauss and Corbin,
Kathy Charmaz is an example of a two-stage process for the grounded theory (Charmaz, 2014).

We used open coding to examine the text for salient categories and labelling passages in the text. Open
coding involves a process towards saturation. We had a constant comparative approach in an attempt to
saturate (theoretical saturation). We therefore were looking for the instances that represent the
category and continue interviewing (looking) until new information did not provide future insight into
the category. With constant comparison, we maintained close connection between categories (the
coding) and data. During the coding process, we categorized concepts that seem to relate to the same
phenomena. This is what is referred to as categorizing.

Interviewing was done in different rounds with the analysis (coding) in between. The idea with coding is
to get away from using the responder’s terminology, to get away from their descriptions to a more
analytical and theoretical terminology. Specific examples from Strauss and Corbin are to use the
terminology “conferring” instead of “talked to a manager” and “information gathering” instead of
“reading the schedule” (Strauss & Corbin, 1998).

Regarding the number of interviews, we found that for phenomenological studies, Creswell (1998)
recommends 5 – 25 and Morse (1994) suggests at least six interviews (Morse, 1994).

Grounded theory aims to be inductive and focuses on generating theoretical ideas (or hypothesis) from
the data. A distinctive feature of the grounded theory approach is the combination of data collection
and analysis simultaneously. The idea is that the researchers collect data and start analysing before

21
finishing the data collection. In grounded theory, data analysis begins to develop theories that suggest
further cases to sample. In other words, the analysis generates the ideas of the sampling. The
theoretical sampling can involve comparison of people, events, setting, places and conditions (Draucker,
et al. 2007). The researcher should try to start with no preconceptions, therefore minimal literature
review should be done prior to the study. The reason for this is that reviewing the literature of the area
under the study is believed to generate preconceptions (Glaser & Strauss, 1967). However, at certain
stages of the analysis you can start to bring in theoretical ideas. There is also a big debate in the
literature about this whether you should involve literature reviews and theoretical ideal. During the
coding process, the naming of the category was created from the 1st order data and not with the
involvement of literature. We had an in vivo coding approach which meant that we placed emphasis on
the actual spoken words of the interviewers (participants). The coding was therefore based on the
participants concepts and their conceptualization. It was therefore directly a form of the understanding
and interpretation of our participants. The categories have multiple perspectives and can be
dimensionalized. As you collect information and perspectives of a particular issue or phenomenon that
have been coded in the same way, slight differences can be observed. In Strauss and Corbin's book, they
talk about the colors as an example. The colors have properties, like shade, tone, intensity and different
dimensions of the properties, like light and dark are different dimensions of shade. In the same way, our
codes can have different dimensions (Strauss & Corbin, 1990). Another example from the book is the
category “watching”, this category has the dimensions duration, frequency, intensity and extent.
Another example is the category “information passing” that can have the properties (dimensions) of
amount of info, or manner of passing.

Grounded theory was originally aimed to discover social and social psychological processes but does not
have to be restricted to this area. Grounded theory can be and has been applied in a whole host of other
areas, as an example in health research areas (Foley & Timonen, 2015). Data is being collected and
analysed before the data collecting is being finished, analytical work is therefore being done alongside
data collection work (Gioia, et al. 2012). Grounded theory is an analytical process that prompts to theory
discovery and development. Theoretical sampling is an approach to preforming the sampling, in
choosing who the researchers are going to interview to get our data. Theoretical sampling is therefore
led by the researchers developing theory, the developing understanding which is related to the
simultaneous process of the data collection and the data analysis. This approach is led by the research
perceptions of the sampling they need to include, based on their background, the kind of setting for the
people they need to collect the data from. The sampling approach is led by the theoretical ideas, the
developing and understanding of the current situation. In grounded theory, data analysis begins to
develop theories (explanations) that suggest further cases to sample (Gioia, et al. 2012). Group concepts
that seem to relate to the same phenomena, are being categorized.

See Figure 3 for more information about grounded theory steps.

22
 Figure 3: Grounded Theory Steps (Gioia, et al. 2012)

3.6. Ethical Considerations

The ethical consideration was presented to the interviewees and the participants of this study in written
form during the confirmation email of the interview but also verbally during the interview process. The
interviewees participation in this study will be confidential. Names and details of the participants will
not be published. The titles and roles of the participants will be presented in the report but no linkage
between the answers and roles and titles was made.

The participants were informed that they were able to withdraw from the interview at any point before
the publication of the study and report. And the participants were informed that they could pass on
answering any of the questions.

The final report will be available to the participants. We asked for permission from the interviewees to
record the interviews for data analysis purposes.

3.7. Validity and Reliability

Regardless of the aforementioned benefits of case study method, the reliability and validity of the
method are challenging to establish but crucial to be present in order for the quality of the study to be
judged as sufficiently objective, rigorous and relevant. Reliability and construct validity are of high
importance in the measurements to prove that they are meaningful and trustable (Ghauri, et al. 2020).

3.7.1. Construct Validity

Construct validity is defined as the degree of which an operationalization measures the concept it
intends to measure (Zaltman et al., 1977). In our study, multiple sources of data collection such as
triangulation of recorded interview videos, documents or public reports and news were used. This
approach is argued by different studies to help increase construct validity (Flick, 1992; Peräkylä, 1977;

23
Hirschman, 1986). Draft case study report should also be reviewed by the interviewees for making clear
no misunderstanding or ambiguity is in place (Yin, 1994).

3.7.2. Internal Validity

Cross-checking of the results during data analysis phase can increase the confidence of internal
coherence of the findings (Yin, 1994). In our study, results were properly cross-checked by different data
sources to verify no inconsistencies present. Besides, clear display of illustrations of analysis process
were visualized to facilitate the explanations of the data analysis of the results, which is argued to have
benefits in making the study more trustworthy (Miles & Huberman, 1994).

3.7.3. External Validity

We have high clarity on the definition of the scope and boundaries in our research question and
research design, which is considered to increase external validity by achieving reasonable analytical
generalizations (Marshall & Rossman, 1989). But generalization of our study is still considered to be
limited due to the nature of single case study.

3.7.4. Reliability
In order to achieve high reliability, there are a number of principles we have strived to be disciplined
with:

- Theories and ideas in our study were extensively described in different phases which can help
increase the rigor (LeCompte & Goetz, 1982).
- All the used materials from data collection will be listed in Appendix section with clear details, in an
effort to prove the data reliable (Lincoln & Guba, 1985). Digital interview data is recorded digitally
and stored properly for review in order not to miss details.
- Peer review was conducted by a peer group for examination (Yin, 1994).
- Clear and executable interview protocol is developed in design phase with a semi-structure (Yin,
1994).
- Pilot-interview was conducted with 3 people unrelated to target interviewees, in an effort to avoid
ambiguity and incompleteness in the questions. This has been argued to improve reliability in
different studies (Yin, 1994; Mitchell, 1993).
- We are two authors with continuous communication and discussion about methodological decisions,
which can reduce individual biases.

24
4. Results and Analysis

In this part, the results of data collection of our case study based on different data sources will be
presented and subsequently analysed in detail.

4.1. Annual Reports

When comparing Thermo Fisher Scientific’s R&D expenses and the company’s revenues over the past six
years, it can be seen that both measures are increasing hand in hand (Figure 4).

1.4 35.0

1.2 30.0

1 25.0

R&D Expenses (billion $)

0.8 20.0

0.6 15.0 Total Revenue (billion $)

0.4 10.0

0.2 5.0

0 0.0
2015 2016 2017 2018 2019 2020

Figure 4: R&D Expenses vs Total Revenue in billion $ (2015-2020) (Thermo Fisher Scientific, 2021)

4.2. Public News and Articles

4.2.1. AI Embedded Solutions for Customers

Thermo Fisher Scientific has launched AI-based processing tools in its Thermo Scientific Amira-Avizo
Software and PerGeos Software which is considered to be a major leap forward in enhancing image
processing capabilities by enabling the cooperation between traditional and AI-based algorithms. By
using deep learning neural networks for processing images, Amira-Avizo Software and PerGeos Software
provide ideal environment that allows researchers to achieve greatly improved results compared to the
traditional image processing (Thermo Fisher Scientific , 2019).

25
On the other hand, for the semiconductor industry, Thermo Fisher Scientific has released commercially
Thermo Scientific Metrios AX which is the world’s first commercial scanning transmission electron
microscope (S/TEM) allowing lab operators to create accurate data promptly with minimum training.
Machine learning technology is the core driver of the automation unit for Metrios AX which simplifies
automation by coordinating all necessarily relevant instructions to identify, align and image the
interesting regions in a sample. This has great potential in saving time and increasing productivity in
failure analysis, process qualification and metrology (BioSpace, 2019). Relevant to this, in 2021 April,
Thermo Fisher Scientific unveiled the Helios 5 EXL Wafer DualBeam which is designed exclusively to
meet increasing sample amount and analysis requests of semiconductor manufacturers. The Helios 5
EXL is including machine learning technology and advanced automation capabilities for precise sample
preparation in semiconductor manufacturing and yield learning, releasing the intense pressure of
semiconductor labs to provide fast transmission electron microscope (TEM) data to support process
monitoring and advance learning curves without cost increases (Thermo Fisher Scientific, 2021).

4.2.2. AGROSAVIA Case

Thermo Fisher Scientific is also at the very forefront of enabling centralized laboratory data
management, which is a key element to facilitate AI in making sure all research data is readily available
with high quality in a centralized location (Thermo Fisher Scientific, 2020). One prominent case is that
AGROSAVIA, a Colombian public corporation for Agriculture Research, was in close cooperation with
Thermo Fisher Scientific for centralized lab data management, Tableau for business intelligence and IBM
WatsonTM for AI capabilities in a project to optimize the soil analysis process by providing farmers with
data-driven recommendations in crop fertilization to achieve increased crop output. AGROSAVIA has
implemented Thermo Fisher Scientific’s software to manage their laboratory networks with centralized
soil data and access to the farming community (Thermo Fisher Scientific, 2020). The interactions
between the different parties are summarized in Figure 5:

26
 Figure 5: Illustration of Interactions in the AGROSAVIA project (Thermo Fisher Scientific, 2020)

In the execution of this project, Thermo Fisher Scientific’s SampleManager LIMSTM (short for Laboratory
Information Management System) software provides massive soil analysis data to IBM Watson TM
intending to utilize its AI capabilities to obtain its model and algorithm for optimal fertilization plans for
farmers. The results of the project have been promising by doubling agronomists’ capability to execute
fertilization plans from 18 to 36 recommendations/day, and have increased farmers’ confidence in
embracing AI (Thermo Fisher Scientific, 2020).

4.2.3. Q&A with CIO of Thermo Fisher Scientific by CIO.com

In our research we have found a highly valuable and informative interview conducted by CIO.com with
the incumbent CIO of Thermo Fisher Scientific, Ryan Snyder (CIO, 2020). A summary of the relevant
answers from Mr. Snyder for our study is as follow.

Adoption to Covid-19: Thermo Fisher Scientific has rapidly and successfully introduced a whole portfolio
of immersive technologies which are transforming how business is done during the Covid-19 pandemic,

27
such as using virtual reality (VR) tools to train their associates from production to sales, conducting
remote customer or regulatory audits by VR, as well as quickly developing 3D models of instruments as
virtual product demos for customers to bring customer experience.

Competencies in leadership needed driving transformation: Mr. Snyder expressed that leaders should
push the boundaries while always connecting innovation with customer values. Besides, empathy and
humility are especially important currently when IT has become the core of the products in product
development and customer experience. As IT capabilities expand beyond the company, a full integration
with R&D, service, and product management with empathy to understand the customer demands and
humility to un-own every IT decision needs to be approached.

An example of IT’s increasing impact on product development: In the past few years, the company has
developed the added complexity of building software and Internet of Things (IoT) sensors into the
products, as well as software which enables customers’ lab operations. The agile teams in digital
engineering organization are helping product teams with delivery.

Technologies considered to have the biggest impact: With cloud services being mature, AI and its
transformational potential are intriguing especially in life sciences. AI is being put into customer service
operations to help customer without service time limitation. AI is also becoming increasing crucial for
gene sequencing and diagnostics in drug manufacturing. Thermo Fisher Scientific has not started the
implementation of AI but by creating hybrid AI teams composed of both IT and product people, the
company can avoid unnecessary reinvention of the whole system. The higher extent of AI capability
modulization, the higher speed for its common use.

The most used leadership skills: Mr. Snyder expressed that the important skills to be used is
empowerment, which gives more decision-making possibilities to the team (CIO, 2020).

4.3. Interviews
As grounded theory was used for our study, all the raw data from the answers of the interviewees have
been coded into 1st order concepts of different categories such as R&D management (People, Process,
Technology and Strategy), R&D leadership and R&D innovation. As a result, the summary of the 1st
Order Concepts from the informants is shown in Table 3 below:

28
Impacts of AI on R&D Coding of Raw Interview Data - 1st Order Concepts

"Inform&Enable" and "People"

-All kind of information can be used for AI. AI provides information about product users, markets and operations via figures and text from
customers through behavior pattern recognition, helping development with safety, ergonomics and simplicity.
- Information sources can come from users using the products, from new research areas about competitors' products including publications
about research, market information.
- Historical information and knowledge can be retrieved deeply by AI retrospectively quickly from internal and external report systems by
researchers for product development.
- Information should be processed by AI similar to data management system to read and categorize information. Funneling information is needed
before it reaches people.
- AI should investigate broadly and compile big data with people's specific aim and should be trained iteratively to understand what the important
areas are, and finally give suggestions based on which decisions are made.
- AI, with clearly defined aims, captures the "pattern" which is the result of its data analysis and shares with people.
- Information receivers can be product designers with hands-on skills, creativity and entrepreneurial mind; both employees at basic levels and as
R&D managers, from all positions from product development, customer-related issues, production deviations, as well as post-marketing (not
limited to R&D).
-Information can come from instruments, software, technologies, processes previous studies and experiments.
-AI output should be validated and evaluated by the employees.
-Processing is needed, to be able to be relevant to employees.
Impacts on People -Plenty of information is available for AI, software can be helpful.
-Information can come from anywhere, this depends on what we are looking for. Could come from customers, surveys from end users and from
the critical mass.
-Difficult to process information, streamlining needs to be considered and what is of interest for us. Need to consider what is possible and
feasible. Data management and processing is much and hard work behind it.
-Information should be processed through a committee composed of different departments. Information should flow in the right order.
-Manufacturing also important and relevant.
-Data for data analysis, but also for customer experience. Different way for getting data.
-Spot-on products, information of how they might want to use the products.
-AI can help with big data for product development.
-Data needs to be visualized in a visual and easy to understand way. Processed data.
-Time is limited for managers and they are exposed to much information.
-AI information can be used to validate the product in the end of development phase. Can also be used to show that the product is safe for
authorities to approve.
-AI information will be used in the beginning of product development phase by employees (to understand which kind of product to develop.
-AI will be used it the end of the product development cycle to understand what to get out of the product.
-AI information can aid during the process, can revisit the different phases.
- AI can use Internal lab data for product improvement to provide analysis and conclusions for lab people.
- AI needs to process data in terms of analyzing and reasoning, as well as verifying the correctness of data.
- Different departments will use the information from AI, such as R&D, production and marketing.
-AI dependent on information that is fed in, and the "teaching".

29
- The source of information depends on what AI is aimed to do.
-The type of data and the kind of data more important than the processing (second layer).
- AI is most applicable in areas with plenty of information, development work is data-intensive. *AI valuable/beneficial in processes related to
data collection and big data.

"Transform" and "People"

- AI requires people to have new skills such as analytical skills on data quality and reliability.
- No new skills or knowledge will hopefully be needed since AI's exceptional capacity of processing large data can assist us, while possibly
simulating outcomes from multiple scenarios for people to select the best option.
- Nothing needs to be changed should remain the same, with the same skillsets required.
- Leaders should understand the needed skills and how to make the changes, in order to plan for employee development and talent acquisition
- Training and education are crucial for people to be onboard in using AI. In people’s nature there is resistance to changes. Positive impact on
motivation, new, exciting and trendy field with much progress. A soft strategy should be used in introducing AI for older generations due to their
skepticisms and resistance against new technology by the threat of AI replacing their jobs.
- The way to "sell" AI to employees is important to avoid demotivation: clear purpose, benefits, and how to succeed.
- General media's coverage on AI, generational preferences, and how the communication is done are important for people's motivation to use AI.
- If the AI used is good, motivation will increase since there is an information overload and this causes frustration in handling the massive data
manually.
- AI gives stronger argumentations with more objectivity for people
- Depending on how AI is set up, knowledge sharing could be decreased from people to people, mainly due to more interactions with AI.
- People need an open mind for AI implementation. People also need more IT and computer skills, to be able to understand the information and
interpret the result.
- Training on AI methodology for gathering information, to be able to be more observant when controlling and validating the result. Instead of
training of the technology related to the specific role, more training will be needed for the AI.
-Component people will still be important (experts in the field), however AI skills would be necessary. If AI will replace tasks related to specific
skills, those skills will not be needed anymore.
-The effect on the motivation will depend on what AI is being used for and employees' interest. Motivation will be mixed, whole spectrum. Will
depend on generation.
-Technical skills of how to use AI needed and skills in data analysis.
-Technically challenging to get people with high skills and knowledge, but training to understand AI data would not be difficult.
-AI’s purpose is to make our lives easier.
-People with different skills for diversity, but not to forget people with experience.
-Creativity and experience meet to make things work.
-AI can make it quicker to find solutions in regards of data sharing in the organization.
-new improvement ideas, and time saving when sharing between groups.
-24/7 sharing can aid knowledge sharing in different geographies.
-Need skills to be able to make use of AI technology (engineering skills).
-Need skills to be able to interpret the AI data. Engineers that have developed the AI needs.
-Long plan is needed for the AI implementation and employee development. R&D is conservative.
-Both managers and followers need to be driving this change to AI.
-AI can stimulate collaboration. Bring different departments together, seeing the bigger picture.
- Better knowledge sharing by bringing better accuracy from massive datasets and detecting some new things that human could miss.
- AI is required in management, it is not self-happening with much work needed. *Purposeful data processing needed for relevance,
interpretation, correctness and information flow direction.
- Knowledge and skills in understanding AI needed, when used in development and in decision making.
- Support and commitment for different ranks and development team need to drive the process and manage data.

30
 - Need the subject matter expertise that need to interpret the data/result and also judge if it makes sense. AI can go wrong, therefore need
someone to control.
- Training, people have confidence in their self and their judgement. Need to understand the potential but also the limitations of AI. Learning
process for the humans. Learn where they can trust it. Need both experience and training.
- Not many changes needed in TA, always need to find talents. Potentially need to find people experienced/knowledgeable with AI.
- Motivation affection will be different, both exited and not. Depends on how AI is programmed and its aim. Like any tool. Bad tools are hated,
and good tools become companions.
- Motivation will be affected positively, can take tedious routine tasks and give people more focus on other certain things more related to human
things.
- Challenge to share information between teams, not positive that AI can help.

"Inform&Enable" and "Process"

- Generally, all the processes in R&D should be included in AI applications, from the very beginning of initial info gathering, to development
phases, tests, until product development is finished.
- Regarding benefits to R&D processes, AI can provide quicker data handling from big data, giving better overview and a more complete picture of
different information, with also possibility in evaluating the outcome probability of different scenarios.
-AI can be applied in both in laboratory work and in management. E.g. Project management, processes, and technology in the lab.
-AI can support onboarding and simplify this, specific learning for different roles/people. Automate certain steps.
-The benefits might be knowledge sharing between different processes in management, more personalized decisions. Saving time moving to the
next step (risk analysis and change request). Processes can be simplified by taking decisions on the next step.
-data collection part to justify the project.
-All processes are important (early development process, validation process, manufacturing, quality assurance)
Impacts on Process -AI is easier to implement in development then in pure research.
-AI collect input requirements, help to process more, more data. Reach more broader then without.
-AI can help track the progress, close down projects that shouldn’t continue.
-AI Benefit is the development products without feelings involved. "someone’s baby",
-AI will influence the decision making. Related to bias when making decisions.
- Lab process done currently by hand can be done by AI, also in reasoning and problem solving.
- AI can collect data and do analysis and give suggestions for people to make decision.
- Faster and more accurate.
- AI can be applied in all areas that have data. In regulated environments old fashioned methods do not accept new methods easy. Slow process.
- Can be implemented easier in internal processes then in external. Can be used in product but Will take time. AI is a black box when used in
products. Why? Need a lot of validation if used in regulated environment.
- Benefits: Data analyzed easier, help in time in processes with a lot of data. AI have to be developed in right way.
- AI is good (superb) with data that is difficult to combine in a traditional means. This is the strength of AI. Better than our traditional methods.

31
 "Transform" and "Process"
- AI provides objective data-based analytics for R&D management in decision making in product design with different information from e.g.
customers, operators and engineers
- From the start of the project, more information can be obtained and better choices can be selected when projects are initiated.
- With the aim to develop products faster with lower costs by using AI in the long run, there can be a shift that products produced in high labor
cost markets would be produced in countries with low labor costs, which leads to a geographic shift of R&D activities.
- AI can change R&D in decision making for the next step. Decision making can be a barrier to execution. AI could suggest examples of decisions
for the employees to choose from.
- AI can aid general tools also, not only related to R&D tools. Travel management, purchasing tools. Could take care of administrative tasks. For
R&D tools, it could aid in understanding the tools with minimal training.
-Quicker and more efficient with AI.
- Easier to make decisions and move forward.
-Helpful in detecting errors and monitoring processes.
-Quality control.
-AI can Influence the decision process. Better analysis and reduce studies. Can Extrapolate better and avoid repeating studies.
-AI can do more simulations, and therefore less tests.
- The whole process will be faster and more accurate with AI, and AI can interpret some information that could be missed by human.
- AI requires cloud-based data management/storage.
-AI can be used in development work, if done correctly by enhancing time.

"Inform&Enable" and "Technology"

- It would be a challenge that AI is smart enough to understand and predict new technology trends.
- Based on big data processing, AI can see pattern and trend. Better overview based on historical data, it is easier for AI to draw conclusion for the
future.
-Provide information from different sources, scientific literature and from different databases. Could for example give suggestions of chemical
structures and bonds between them.
- -We can streamline what we are interested in by using AI by filtering the noise.
-AI can simply things by e.g. adding new possibilities or eliminating things that do not give added values.
-It can be used to predict new technology. Maybe more then we think. It can pick up trends quicker.
- AI has better predictability than human in some area, such as drug development.
Impacts on Technology - - Do not see how AI can understand and predict technology.

"Transform" and "Technology"

- AI can possibly bring changes in the way R&D invests new technology to produce new products, while AI replaces physical lab and reduces
operational and labor costs.
-Could suggest new technologies for people to decide on and would bring more options with a wider view.
- R&D will see the opportunities and benefits with AI. R&D will move forward because AI gives opportunity for easier access and easier ways of
working in the future. There will be interest in investing in this area.
-AI can predict or be one way of predicting, same methods will probably still be used but AI will not involve emotions in the process.
- AI can change things, will affect other industries as well. Slower in the industry Thermo is working in. Need a something that will initiate the
change, some macro factor.

32
 "Inform&Enable" and "Strategy"
- AI contributes to customer centric innovation by informing R&D management how users use products, how to provide them with faster and
safer solutions, how to train customers in using the product and give suitable suggestions to customers to help their workflows.
- AI can contribute to user data of competitors' production, and do the trend analysis for seeking future opportunities.
- AI can help us be quicker to get new products in the high-growth market and help customers in need quicker. AI can spot out the right trend in
emerging markets to develop products for.
- AI contributes to delivering unique value proposition to customers by making more personalized products with more information on customer
needs and their usage of products, and more understanding on customers' pain points.
- AI contributes to suggesting the future trends of the need of massive population, not just individuals.
- AI contributes to quicker adaptation of customer needs and quicker delivery to customers.
-Involve the customers and looking at it from customers point of view, looking at their needs.
-AI would look into ways to create this innovation based on customer needs.
-AI could speed up the process and get faster decisions, based on e.g. different regulations in different markets (EM and HG markets). Can quicker
interpret the regulations.
-Use AI to inform customers of our offerings, and how it would benefit them.
Impacts on Strategy -High grow markets, new things and technologies are taken in a faster pace. Early adopters. Keen to jump into using new technology to simplify
and make things easier.
-Slower adaptation in emerging markets.
- AI can provide new value proposition, unique selling points by showcasing them to the customers
-AI can help evaluate and simulate customers and get the bigger user base. It is costly to go out to customers.
- AI can help with the analysis of market needs, marketing analysis about the customer needs.
-AI could also accelerate development and market access of a product
-EM are difficult to get (our) products in to. If could get a better understanding of what we need there, it could accelerate growth.
-AI could help to develop the value proposition, to make that more powerful.
- AI improves product with higher speed and accuracy, making customer more satisfied.
- New ideas can be tested faster
- AI contribute to customer centric innovation and contribute to the growth and emerging markets by helping developing new products. Faster
launch of products. Can speed up.
- When much data, tedious to do the development. Speed things up.
- Help to reduce manual work. Making decisions.

"Transform" and "Strategy"

33
- Depending on the strategic role of AI the company is positioning, AI provides decision makers with stronger ground and more informative and
reliable suggestions in their decision making.
- AI can minimize administrative burden and people have more time to work on practical activities.
- AI can analyze big data to come up with suggestions on management decisions on prices, locations and timing to produce.
- AI can influence revenue positively by providing R&D with faster decisions and adapting customers’ needs quicker, and possibility to shift faster
before the real shift occurs, as well as by reducing costs and increasing quality.
- AI can increase revenue by the reduction of man hour cost put in the development of new product and delivery to customers. Fewer people will
be needed in R&D in the lab.
- AI can provide stronger evidence on which investment and how investment should be made. R&D investment can go up by investing in AI and
integration to different systems.
- AI can decrease the monetary need in investment due to the decrease of manpower cost which is the highest cost in R&D
- AI can replace some manual work and ways of doing things. But people should have same roles with extended assistance of AI with faster and
more accurate information.
- Organizational hierarchy can be affected by AI to be flatter because there is possibility that some middle layers' tasks could be replaced by AI.
-Predict the resources needed, based on the projects that are in pipeline.
-Can help allocating resources between groups.
-Efficient prioritization of the work for the chosen projects.
-Can both increase and decrease the budget, depending on AIs interpretation.
- Could potentially flatten the organization structure, with less levels.
-Data that is being captures and presented to higher management.
-AI can showcase the summarized information to GLT or decision makers, which makes it easier for managers to explain the logics behind. We do
not need to show all the data points but only the important ones which AI helps you in identifying.
- AI can interpret information for management from massive data.
- AI can help revenue increase in different positions: sales, new product development, etc. AI can close the information loop by feeding back
information from one end to another end. In this way it can help bring more revenue to the company.
- When you need to make a new investment, it will need massive data and justify all the data which AI can provide its help with, in an
understandable way. The database is big and the significance of the data will be high and validity is therefore high.
-A lot of processes will be changed, and will be easier.
- Maybe certain types of job will not be relevant for the future, meaning certain roles will be replaced by AI.
- It will not happen overnight but gradually. Companies should educate people in the areas which can be replaced by AI to work in other areas so
that they can work in somewhere different.
-AI in decision making will be a factor in the future, AI will support the decision but providing information and analysis.
-AI will not be able to replace us, human touch. But will be a factor we take into consideration.
-AI will get more out of the organization. By using good tools, with AI and other tools, and drive efficiency.
-R&D investment will take longer time to implement AI in.
-AI will influence certain things in the org structure, e.g. service centers (complaints and routine things).
-AI could replace people at certain levels, but more "difficult" things need to be handled by people.
-Data is already available but not really accessible for R&D (from marketing e.g.). AI can help here.
- AI helps visualize workload in different groups and hence improve decision making in resource allocation.
- Good for revenue increase by non-stop working around the clock and faster and more accurate analysis.
- Decisions need to come from the corporate management. Will require resources to manage the data.
- Being able to launch products faster and will have more products.
- Can influence revenue if a lot data from consumers is available. Difficult to use AI in data poor companies (compared to google, FB and other).
- AI can save money if replaces employees.
- In future, experts like doctors and lawyers, will be accompanied with different AI devices. Taking over routine tasks. AI will reduce our need.
Less need for certain jobs, like less doctors needed because more effective with AI. More effective.
- When judgment is needed, humans are better. Other certain (admin related) task AI could help.

34
 AI Impacts on R&D Leadership

- AI can help assist leaders in product development with speed and safety
- There is an imminent need for leaders' agility to start getting understanding on how AI can help them, which skills are needed.
- AI provides a way for leaders to get information much more quickly and get a better overview from the enormous data access with AI's
exceptional capability to process and analyze data. Leaders will have lower risks of missing important details or information from the massive
Leadership Implications data which AI has access to. Research publications and new regulations are being updated every day and it is impossible for people to follow-up,
and AI can help with these with high objectivity, accuracy and speed.
-Decisions will be taken without the involvement of emotions.
-Saving time and bring more efficiency
-Have not seen it so far in my leadership but believes it will come, in the future. Will be an additional tool for decision making in the future.
- Making process faster and more accurate, and replace labor cost on some manual work.
- Do not want and let it affect my leadership at all. Leadership requires social skills, AI not able to replace this
- Could potentially affect leadership but current AI is far away from that. Routine work can be replaced.

How to Achieve Successful R&D innovation with AI

- Specific support, sufficient investment in technology and people who are open for changes, where change management is crucially pivotal
throughout the organization.
- Technically, AI needs to have the capability to give as much background information as possible and simulate the outcomes of different
scenarios, and then curate the information from huge amount of data to provide suggestions to help people make decisions.
- AI should be used in many ways, not only for product development, but also in other functions such as post-marketing to understand how well
the product is received by market and also market share.
Innovation Implications - The most challenging thing for getting success in using AI can be to find the best AI tool and get it working. The big difficulties lie in maintaining
it and get it working as it is desired to be in implementation.
-AI can streamline the R&D and innovation.
R&D instrument, evaluate the footprint before the actual product. AI can aid.
- AI can help indicate if to move forward with a R&D project.
-The analysis part, tools that can draw conclusions quicker then we can today. Where I am working in, it is mainly data analysis. Believe that this
will differ in other areas.
- It should be clearly communicated that AI is just a digital assistant that makes jobs easier and would not incur unemployment.
- Need to have the correct ingredients (data, and understanding the mixing of ingredients).
- Leadership team needs to be committed, and understand that it is not an easy process. Different type of thinking.

Table 3: Coding of Interview Data – 1st Order Concepts summary

35
4.4. Analyses of the Results
Based on the results presented in 4.1-4.3, further analyses are done and presented in the following
parts.

4.4.1. Analysis of Secondary Data – Annual Reports, Public Articles and News
From the annual reports, we can see the clear upward trend in the R&D expenses and the company’s
total revenue in the past few years, apart from a sharper increase during 2020 due to the Covid-19
pandemic (Figure 4). This is in consistence with the company’s growth and innovation-driven strategy
and these two measures are expected to continue to increase according to the strategy. It is argued
that a company’s commitment and investment spent on R&D contributes to the firm’s financial
performance (Mansfield, 1998). Especially for innovation-driven enterprises, R&D expenses has
proven to have a positive correlation with the company’s financial performance (Rijanto, 2018).
Therefore, we consider the continuous commitment and substantial investment in R&D can
continuously be the two of the drivers for Thermo Fisher Scientific’s increase in profit and growth.

The examples in 4.2 of AI-based software solutions for customers in image processing and
semiconductor sample preparation, including the AGROSAVIA case, reflect that Thermo Fisher
Scientific is aware of the increasing customer demands for faster and more accurate solutions and
the capabilities that AI is able to bring for fulfilling these demands. This is also a strong indication of
the firm’s increasing emphasis in the development of AI-based solutions in its products to create
extra added values for customers.

The answers from the firm’s CIO’s interview provide highly valuable insights from the top
management’s perspective for our study. We understand that AI is clearly considered as “game
changer” for the future due to its transformational capabilities in different areas, although Thermo
Fisher Scientific has not yet taken the big step to initiate AI implementation. However, AI
implementation is a must-take step for the company to achieve future success. While IT technologies,
together with AI, become the core of products, an organizational integration of different
departments is foreseen which is a result of IT capabilities expanding beyond the firm where people
from different functions of the organization need to work together and empathize customers’ needs.
In the future, it is foreseeable that with more AI applications after the implementation there will be
certain decisions which leaders should let go and let the IT and AI technology make its decision. This
will bring the shift of some of leadership elements, such as a reduction on reliance on personal
expertise and cognitive intellect.

But, serving as a macro factor, Covid-19 pandemic has driven leaders to push the boundaries in
innovations in different areas, such as production and sales as well as customer service, to
successfully adopt to the mega-change in the global environment while staying connected with
customer values. The key messages are that leadership will still be crucial to succeed in
transformation, where empathy and humility are considered to be the necessary elements leaders
should be equipped with. At the same time, empowerment should be emphasized in leadership to
facilitate employee motivation and innovation, which is also positively argued by Cromb (2007).

36
4.4.2. Analysis of Primary Data – Interviews
In this part, a deep and thorough analysis is presented based on primary interview data, where 2nd
order themes and further the aggregated dimensions are distilled based on the aforementioned 1st
order concepts.

4.4.2.1. From 1st Order Concepts to 2nd Order Themes

From the above results summarized in the Table 3, all the 1st Order Concepts in each dimension are
further elevated abstractly onto the 2nd order theoretical level of themes according to their
similarities and differences, which attempts to answer the important question “what is going on
here?” (Gioia, et al. 2012). As interviews were carried out one after one, we saw a clear propensity
that the 1st order concepts became increasingly saturated with less and less new ideas created
through constant comparison, which makes the 2nd order themes denser with the 1st order inputs
and conceptually saturated as well.

As one example, the 1st order concepts “All kind of information can be used for AI. AI provides
information about product users, markets and operations via figures and text from customers
through behavior pattern recognition, helping development with safety, ergonomics and simplicity”,
“Information sources can come from users using the products, from new research areas about
competitors' products including publications about research, market information”, “Information can
come from instruments, software, technologies, processes previous studies and experiments”,
“Historical information and knowledge can be retrieved deeply by AI retrospectively quickly from
internal and external report systems by researchers for product development” and “Provide
information from different sources, scientific literature and from different databases. Could for
example give suggestions of chemical structures and bonds between them” are elevated into the 2nd
order theme as: “Quicker and broader access to internal and external information (customers,
market, competitors, operations, historical data and scientific research)”. By using this same
approach, we have abstracted all the 1st order concepts into totally 21 respective 2nd Order Themes
as below (Table 4):

2nd Order Themes

1 Quicker and broader access to internal and external information (customers, market,
competitors, operations, historical data and scientific research)
2 Simpler and more objective in analysis and more efficient decision making
3 Purposeful data processing needed for ethics, relevance, interpretation, correctness and
information flow direction.
4 AI information generally applicable to managers and employees at lower levels
5 Questionable change in the required skillsets for people
6 Necessary trainings in usability (IT skills), interpretation (statistics) and change management

7 Updated employee development plan and talent acquisition strategy for managers
8 Good motivator for employee if communication is well and AI is practically useful
9 More knowledge sharing and collaboration with different departments in most cases

37
10 AI to be implemented in different processes of R&D.
11 Necessary human control in AI outputs
12 AI valuable/beneficial in processes related to data collection and big data.
13 Streamlining and simplifying processes.
14 Better predictability and provision of multi-scenario suggestions to human
15 More customer-oriented and better value proposition by more efficient customer interactions
and more precise understanding of customer needs
16 Quicker growth in high-growth and emerging markets with better intelligence in demand,
competitors and customer behavioral patterns.
17 Replacing certain human elements (e.g hard skills) and flatter organizational hierarchy

18 Higher efficiency (quicker/more accurate/ cost)

19 Guaranteed management commitment and understanding of AI limitations for innovation
success
20 Sufficient and correct technology investments
21 Big challenges and needs in finding the most suitable tool and implementing it
Table 4: 2nd Order Themes summary

The detailed processes of 1st order concept distillation into 2nd order themes are summarized in
Appendix 3.

4.4.2.2. From 2nd Order Themes to Aggregated Dimensions

After getting the workable set of themes and concepts we have achieved what is termed as
“theoretical saturation” (Glaser & Strauss, 1967). Based on this, we strived to continue to work on
the distillation of the 2nd order themes to what is termed as “aggregated dimensions” (Gioia, et al.
2012). In this section, we will describe the process and motivation of distilling the 2nd order themes
to aggregate dimensions. As summarized in 4.4.2.1, we identified 21 themes in the process of 2nd
order coding. Those 21 themes were later coded into 8 different aggregate dimensions that we
discovered as the following diagram (Figure 6) shows.

38
 2nd Order Themes Aggregated Dimensions 2nd Order Themes

1.Quicker and broader access to

internal and external information
(customers, market, competitors,
11. Necessary human control in AI
operations, historical data and
outputs
scientific research)

2. Simpler and more objective in

1. Shift in leadership 12. AI valuable/beneficial in
analysis and more efficient decision elements processes related to data collection
making and big data.

13. Streamlining and simplifying

3. Purposeful data processing
processes.
needed for ethics, relevance, 2. Shift in employees'
interpretation, correctness and (managers and followers)
information flow direction
needed skills
14. Better predictability and
provision of multi-scenario
suggestions to human

4. AI information generally 3. Organizational 15. More customer-oriented and

applicable to managers and integration (streamlining & better value proposition by more
employees at lower levels flattening&less silo) efficient customer interactions and
more precise understanding of
customer needs

16. Quicker growth in high-growth

4. Management strategy
and emerging markets with better
5. Questionable change in the forming aid intelligence in demand, competitors
required skillsets for people
and customer behavioral patterns.

6. Neccesary trainings in usability 17. Replacing certain human

(IT skills), interpretation (statistics) 5. Objectivity by avoiding elements (e.g hard skills) and flatter
and change management emotions and biases organizational hierarchy

18. Higher efficiency (quicker/more

7. Updated employee development
accurate/cost)
plan and talent acquisition strategy
for managers
6. Efficiency booster
19. Guaranteed management
commitment and understanding of
8. Good motivator for employee if AI limitations for innovation
communication is well and AI is success
practically useful

7. Innovation stimulator
20. Sufficient and correct
9. More knowledge sharing and technology investments
collaboration with different
departments in most cases

8. Profit and growth driver 21. Big challenges and needs in

finding the most suitable tool and
10. AI to be implemented in
implementing it
different processes of R&D.

Figure 6: From 2nd Order Themes to Aggregated Dimensions

Aggregated Dimension 1: Shift in leadership element

Theme 2 (Simpler and more objective in analysis and more efficient decision making) is linked to the
1st aggregate dimension “shifts in leadership element”, as this is related to decision making and
analysis and having an objective approach. Theme 6 (Necessary trainings in usability (IT skills),
interpretation (statistics) and change management) is also related to this dimension relating to not
only the shift in leadership element with the new skills needed in usability and interpretation, but
also in change management. In the same sense, Theme 7 (Updated employee development plan and
talent acquisition strategy for managers) is connected to this dimension where an updated employee
development plan or a talent acquisition strategy is highly relevant for the shift in leadership
elements. Theme 11(Necessary human control in AI outputs) in having a control step for humans in
the AI output, will require a shift in the leadership element for leaders. An example of leadership

39
elements that are believed to be needed more is skills related to humility, adaptability and empathy,
which are highly coherent with what the company’s CIO was suggesting in the public interview report
(CIO, 2020). This could further be related to soft and hard skills. It is also believed that some tasks
could be performed by AI making the need for some of the skills of employees and managers
redundant. This relates to the last Theme 17(Replacing certain human elements (e.g., hard skills) and
flatter organizational hierarchy), where the perception is that certain leadership elements will be
replaced by AI.

Aggregated Dimension 2: Shift in employees’ (managers and followers) needed skills.

We got contradicting perceptions on the required skillset for people in Theme 5 (Questionable
change in the required skillsets for people) which relates to the second aggregate dimension. One of
the perceptions was that new skillsets are needed to be able to cope with the implementation of AI,
while others believed that no skills would be needed. “AI’s purpose is to make our lives easier.” is a
citation from one of the interviews. We interpret that some kind of initial investment could be
needed in new skills to be able to reach the stage where AI is simplifying things for humans.

If AI was developed in an optimal way, it would be able to replace human intelligence and would
therefore not require new skillset. Another understanding was that certain skills in the form of
analytical skills would be needed, to be able to interpret the analyzed data and the processed data.
Skills for the development of the AI is a skillset that is believed to be needed. Theme 6 (Necessary
trainings in usability (IT skills), interpretation (statistics) and change management) was coded into
this dimension and was believed to be needed for the successful implementation of AI. Change
management from the same theme (Theme 6) also relates to the shift in needed skills, and this is for
the leadership team to be able to cope with potential resistance from followers related to the
changes that would take place. Theme 2 (Simpler and more objective in analysis and more efficient
decision making) links to the shift in needed skills and therefore to this aggregate dimension. The
idea is that AI would replace some of the elements, if not all, therefore the needed skills for both the
managements and followers should change.

Aggregated Dimension 3: Organizational integration (streamlining & flattening & less silo)

Theme 9 (More knowledge sharing and collaboration with different departments in most cases) was
coded to the aggregated dimension of “organizational integration (streamlining & flattening & less
silo)”. The idea here is that AI would aid the knowledge sharing in the organisation in the way that it
can both directly replace sharing between employees but also enable quicker finding solutions in
regards of data sharing in the organization. AI would also be able to share information 24/7 in
contrast to sharing between people which is limited to the availability of the people and dependent
on the time zone and the different geographies. AI is also believed to stimulate collaborations by
bringing different departments together, and thus “seeing the bigger picture”. More knowledge
sharing within the organization will not only affect the integration of the organization, but also
relates to streamlining and flattening. Theme 4 (AI information generally applicable to managers and
employees at lower levels) also relates to this dimension, and one assumption is that AI will be
applicable and usable by both managers and followers and that it is not necessarily limited to one or
the other. Theme 17 (Replacing certain human elements (e.g hard skills) and flatter organizational

40
hierarchy) was also coded to this dimension. One assumption here is that AI can replace some
manual work and ways of doing things. But people should have same roles with extended assistance
of AI with faster and more accurate information. It is believed that organizational hierarchy can be
affected by AI to be flatter because there is possibility that some middle layers' tasks could be
replaced by AI.

Aggregated Dimension 4: Management Strategy Forming Aid

This dimension was coded from Theme 2 (Simpler and more objective in analysis and more efficient
decision making), Theme 7 (Updated employee development plan and talent acquisition strategy for
managers) and Theme 14 (Better predictability and provision of multi-scenario suggestions to human)
relate to the management strategy forming aid that we have identified. The belief here is that AI can
simplify the analysis in terms of processing the data quicker, meanwhile having an objective
approach. It is believed that AI will make processes faster and more accurate and replace labor cost
on some manual work. AI in contrast to human is assumed to base decisions without biases and
favors. In other words, it is believed that AI will be able to take decisions without the involvement of
emotions.

Having a more simple and objective process in the analysis will aid the updated employee
development plan and talent acquisition strategy for managers (Theme 7). It is considered that
leaders should understand the needed skills and how to make the changes, in order to plan for
employee development and talent acquisition. One perception was that long plans are needed for
the AI implementation and employee development, as R&D is a conservative domain. Another
perception was that not many changes are needed in talent acquisition since there is always a need
and challenge to find talents. Potentially, talent acquisition has a need to find people
experienced/knowledgeable in AI in addition to subject matter expertise.

Also, the perception to have better predictability and provision of multi-scenario suggestions to
human is also related to this dimension as a management strategy forming aid (Theme 14). One of
the beliefs was that based on big data processing, AI would be able to see patterns and trends and
would have a better overview based on historical data, and that it is easier for AI to draw conclusion
for the future. AI is also perceived to be able to filter noise and thus give the possibility for the
employees and managers to focus on what is relevant (interesting). AI could also simplify by e.g.,
adding new possibilities or eliminating things that do not give added values. We got some
contradicting perceptions, where one of the interviewees believed that they cannot see how AI can
understand and predict technology.

Aggregated Dimension 5: Objectivity by Avoiding Emotions and Biases

This dimension was distilled from Theme 2 (Simpler and more objective in analysis and more efficient
decision making), Theme 3(Purposeful data processing needed for ethics, relevance, interpretation,
correctness and information flow direction), and Theme 4(AI information generally applicable to
managers and employees at lower levels), which is another prominent impact of AI bringing higher
level of objectivity in information by avoiding human’s emotional implications and cognitive biases.
This is highly relevant when we have not only AI’s purposeful data processing based on right aims in

41
Theme 2 which increases the comprehensiveness, integrity and correctness of the data, but also the
general applicability to all levels of employees in Theme 3 which increases the transparency of the
data. All in all, this can help avoid the cognitive biases based on people’s emotions, experience or
cognition to make efficient decisions, which is manifested also in Theme 4.

Aggregated Dimension 6: Efficiency Booster

For this aggregated dimension, it is abstracted from Theme 1(Quicker and broader access to internal
and external information (customers, market, competitors, operations, historical data and scientific
research)), Theme 2(Simpler and more objective in analysis and more efficient decision making),
Theme 3(Purposeful data processing needed for ethics, relevance, interpretation, correctness and
information flow direction), Theme 4(AI information generally applicable to managers and employees
at lower levels), Theme 8(Good motivator for employee if communication is well and AI is practically
useful), Theme 10(AI to be implemented in different processes of R&D), Theme 12(AI
valuable/beneficial in processes related to data collection and big data) and Theme 18(Higher
efficiency (quicker/more accurate/cost)). The word “efficiency” has multiple layers of meaning
including efficiencies in internal and external communication for R&D, decision making, knowledge
sharing within department and across departments, product development processes, problem
solving and cost, etc. AI provides much broader and faster access to different kinds of data to all
levels of employees (Theme 1&4&12) with purposeful processing (Theme 2) which brings accuracy
and objectivity. Analysis and decision making will be faster than before (Theme 3). If communication
of AI introduction and proper change management are done well, people would be more convinced
AI is a digital assistant in helping out their daily work and making their lives easier, which can bring
up the motivation at work (Theme8). By partially replacing some manual work, such as routine
hands-on laboratory work and administrative and analysis work, more man hours would be released
for more value-added tasks or there can be possibility for optimizing headcounts which reduces
labour cost which is considered as one of the biggest costs in R&D. As above, AI plays a convincing
role as Efficiency Booster in R&D.

Aggregated Dimension 7: Innovation Stimulator

This aggregated dimension is extracted from Theme 1(Quicker and broader access to internal and
external information (customers, market, competitors, operations, historical data and scientific
research)), Theme 8(Good motivator for employee if communication is well and AI is practically
useful), Theme 9(More knowledge sharing and collaboration with different departments in most
cases), Theme 10(AI to be implemented in different processes of R&D), Theme 19(Guaranteed
management commitment and understanding of AI limitations for innovation success), Theme
20(Sufficient and correct technology investments) and Theme 21(Big challenges and needs in finding
the most suitable tool and implementing it). Innovation is simulated by multiple factors and based on
some pivotal premises. The prompt and comprehensive access of internal and external data (Theme
1) is capable of providing valuable and convincing inputs with timeliness, accuracy and objectivity for
creating new ideas in product development. With better employee motivation (Theme 8), improved
internal and interdepartmental knowledge sharing (Theme 9) as well as AI implementation in every
process of R&D enabling more streamlined and efficient processes (Theme 10), novel ideas are easier
to be created, shared and further implemented more efficiently, which stimulates innovation. At the

42
same time, substantial management understanding on AI limitations and strong commitment for
implementation (Theme 19) are an important prerequisite for innovation. Moreover, investments on
related AI technologies should be sufficient (Theme 20) and a considerable magnitude of efforts
need to be dedicated to searching or developing the most suitable tool and its implementation.
These are important premises to ensure innovation success in R&D when using AI.

Aggregated Dimension 8: Profit and Growth Driver

This dimension is induced from Theme 15(More customer-oriented and better value proposition by
more efficient customer interactions and more precise understanding of customer needs), Theme
16(Quicker growth in high-growth and emerging markets with better intelligence in demand,
competitors and customer behavioral patterns) and Theme 18(Higher efficiency (quicker/more
accurate/cost)). Thermo Fisher Scientific has three strategic pillars which are “Customer-centric
Innovation”, “Delivering a Unique Value Proposition” and “Leveraging Scale in High Growth and
Emerging Markets”. The first two pillars are included in Theme 15 which has an exceptional emphasis
on understanding and fulfilling customer needs, while Theme 16 is represented as the third pillar. AI
is conceived to have a strong power supporting these three strategic pillars by acquiring, analyzing
and even learning from massive data from different sources, such as social media feeds, feedback
from intelligent sensors or analytics from external databases, in order to provide decision makers up-
to-date understanding and comprehensive insights from the customers’ demands, to further develop
products that can create important added values for customers, and finally to fulfil the growth in the
high-growth and emerging markets by making customers satisfied in different geographics. This is a
strong driver for the profit and growth of the company. Besides, the higher efficiency in different
dimensions (Theme 18) summarized in Aggregated Dimension 6, helps cost reductions and
innovation delivery acceleration, which is also giving positive contribution to profit by reduced costs
and growth via faster innovations delivered to customers.

The detailed distillation process from 2nd order themes to 8 aggregated dimensions summarized
above is also presented in Appendix 4.

4.4.2.3. Theoretical Proposition – Dynamic Model Formulation

With the full set of 1st order concepts, 2nd order themes and the aggregated dimensions, as well as
the complex interrelations between them visualized in Appendix 3 and 4, the whole “static picture”
of the phenomenon of how AI impacts management and leadership in R&D is present. However, the
grounded theory should not only contain the static elements but also include the visualization of the
dynamic relationships among all the concepts to clarify the data-to-theory linkages (Gioia, et al.
2012). Therefore, a dynamic theoretical model is formulated, as Figure 7, with the 2nd order themes
and the aggregated dimensions, as well as the arrows for describing the dynamic mechanism since
arrows are considered to be capable of setting everything in motion (Nag, et al. 2007).

43
 Step 1 Step 2 Step 3

1. Shift in leadership
elements
Implementation of AI application in R&D

1. Quicker and broader

6. Neccesary trainings in access to internal and
11. Necessary human usability (IT skills), external information
interpretation (statistics) (customers, market,
and change management competitors, operations,
2. Shift in employees' historical data and scientific
research)
(managers and
2. Simpler and more
followers) needed objective in analysis and
8. Good motivator for
more efficient decision
skills employee if communication
is well and AI is practically

6. Efficiency
10. AI to be implemented in
17. Replacing certain human
elements (e.g hard skills) and
booster different processes of R&D.

7. Updated employee flatter organizational hierarchy

development plan and 3. Organizational
talent acquisition strategy
12. AI valuable/beneficial
8. Profit and
for managers
integration in processes related to data
collection and big data. growth driver
(streamlining & 9. More knowledge sharing
and collaboration with
flattening&less silo) different departments in most 18. Higher efficiency
cases
7. Innovation (quicker/more accurate/
cost)
stimulator
4. AI information generally 19. Guaranteed
4. Management applicable to managers and management commitment
and understanding of AI
5. Questionable change in limitations for innovation
the required skillsets for strategy forming aid
people
3. Purposeful data 20. Sufficient and correct
processing needed for ethics, technology investments
relevance, interpretation,
correctness and information
flow direction
21. Big challenges and
needs in finding the most
5. Objectivity by suitable tool and

avoiding emotions 15. More customer-oriented

and better value proposition
and biases by more efficient customer
interactions and more precise
16. Quicker growth in high-
growth and emerging markets
understanding of customer with better intelligence in
needs demand, competitors and
customer behavioral patterns.

Figure 7: Formulated Grounded Theory – Dynamic Model

As Figure 7 shows, after implementation of AI applications is initiated in R&D, aggregated dimensions

of impacts such as shift in leadership elements which is highly consistent with section 4.4.1, shift in
employees’ needed skills, organizational integration, management strategy forming aid and
objectivity by avoiding emotions and biases will be triggered as Step 1. All the impacts mentioned
above proceed to Step 2 to further impact efficiency and innovation, eventually leading to driving
profit and growth (via Step 3).

44
5. Discussions

In this part the answers to the two formulated research questions will be discussed in detail.
Furthermore, main limitations of this study and implications for future researchers and practitioners
will also be discussed.

5.1. Answers to Research Questions

Thermo Fisher Scientific is an innovation-driven company with a clear mission “to enable our
customers to make the world healthier, cleaner and safer”. It is evident from the annual reports and
public news that the company has a solid commitment to innovation through its continuously
increasing R&D investment throughout the years and also has an increasing focus in development of
AI embedded solutions for customers and central data management working closely with AI, which
manifests a clear agility and ambition for AI development and application in this global mega-trend.

Thanks to the exhaustive analyses of the collected data from the interviews and the secondary data
sources, this study has provided insightful and interesting answers to both of our formulated
research questions.

For Research Question 1 “How do AI applications impact on management and leadership in R&D
department in Thermo Fisher Scientific?”, as for management, our analyses based on the interviews
first shed light on impacts of AI application on R&D management in Thermo Fisher Scientific. Firstly,
upon AI implementation, a shift of skillsets that is needed for both managers and followers is
expected to emerge due to the adoption of the novel technologies. Skills such as analytical skills on
data quality and reliability, as well as how to use AI for analysis are argued to be required for
successful implementation, which is coherent with the findings from Sjödin et al. (2018) that people
need to be educated to develop the competence to utilize connected data systems. When
considering new recruitments, managers need to think through the new requirements in talent
acquisition. Also, AI application is expected to facilitate organizational integration by streamlining the
R&D processes and breaking down the silos with other departments with reinforced internal and
inter-departmental knowledge sharing. This impact is also pointed out in our secondary data analysis
from the interview with Thermo Fisher Scientific’s CIO Ryan Snyder by CIO.com that a full integration
between R&D, service and product management are foreseen as IT capabilities expand beyond the
company to reach customers (CIO, 2020). Moreover, with AI application implemented in R&D, our
primary data analyses point out that management strategy formulation in R&D, such as resource and
investment planning and project prioritization, will be strongly bolstered by AI through getting
information and suggestions regarding different areas such as market shares for products, changing
customer needs, competitor developments, latest technologies from research, updated regulations
and scenario simulation for product designs with timeliness, accuracy, comprehensiveness,
objectivity and less risks due to human emotions and perceptional biases. This is coherently
resonated by Xing et al. (2020) that AI-based R&D can bring advantages like more objective
consumer requirements and more precise market trend prediction, as well as lower risk in R&D
processes. All in all, a strong consensus from the interviews is that higher efficiency will be achieved
in R&D processes with AI application implemented with all the advantages discussed above in terms

45
of people, process, technology and strategy, which coheres with the findings in multiple studies (Xing,
et al. 2020; Blackburn, et al. 2017; Farrington & Alizadeh, 2017; Sjödin, et al. 2018).

In terms of leadership, the analyses of the interviews further suggest that upon the implementation
of AI application in R&D department in Thermo Fisher Scientific, R&D leaders are expected to see a
shift of their leadership elements which become less focused on the “hard” parts based on AI’s
provision of up-to-date, correct, objective and insightful information and analyses through e.g.
pattern recognition from consumer behaviour and market demand or virtual experimentation and
simulation, for leaders to make decisions, which would consequently partly substitute the
traditionally highly-recognized “hard” leadership skills such as cognitive ability, expertise, experience
and decisiveness, meanwhile releasing more time for leaders to focus more on the value-added
activities. This impact is highly consistent with the findings by Chamorro-Premuzic et al. (2018) that
the essence of effective leadership should be redefined due to the fact that “hard” elements such as
profound specialized expertise, authority and decisiveness and short-term focus on tasking can step-
by-step be replaced by AI, whereas the “soft” elements like humility, adaptability, vision and
engagement will become the critical elements for leadership. In this respect, Huang et al. (2019) also
echo in their study that AI has strong advantage in mechanical intelligence and thinking intelligence,
but is rather weak in feeling intelligence which refers to a leader’s capability to recognize, emulate
and respond to human emotions appropriately. Furthermore, the same coherence is found in the
external interview data with the company’s CIO, where the CIO expressed the criticality of
implementing AI for the company’s future and leaders should delegate partly decision making to AI
while having strong focus on empathy, humility and empowerment (CIO, 2020).

As for our Research Question 2 “How can AI applications be utilized to stimulate innovation in R&D
department in Thermo Fisher Scientific?”, our study has also many interesting findings for answering
the ways AI applications that can stimulate innovation in the R&D.

In general, the perception from the interviewees was positive on this aspect despite that there were
also some challenges pointed out. The idea is that with the help of AI, analysis process can be both
be simplified and done in a broader extent and thus boost innovation in the organization. It is
believed that conclusions can be drawn quicker, especially when the decisions are dependent on the
processing of data and data analysis. The belief is that AI can be replacing some of the tasks that
requires hard skills and thus the followers and managers are enabled to focus on other skills more
related to innovation. Administrative work is one example of tasks that are believed to be replaced
by AI. Simple administrative tasks can be replaced by computers (without AI), however there are
some administrative tasks that still need to have the involvement of human intelligence, which
would be the target for AI to replace. AI is therefore not believed (with the current AI technology,
ANI) to directly stimulate innovation, but by enabling sophisticated data analysis and replacing the
hard skills, AI will be able to enable and promote the employees to increase innovation in the
organization. This also goes along with the literature review we found in this area (Chamorro-
Premuzic, et al., 2018), suggesting a shift in leadership elements for the management, where AI is
potentially able to replace the hard skills. This also relates to the higher efficiency and the
streamlining processes that AI would bring, and indirectly enabling innovation.

46
There were also challenges identified in utilizing AI applications to stimulate innovation. One
challenge that was brough up by multiple interviewees was the resistance to change by the
employees, this was especially relevant in R&D departments that was described as being
“conservative”. Change management therefore was identified as critical for addressing this
resistance, together with a commitment leadership team and clear communication with the actual
impact on the employees. Technical challenges were also something that was brough up by the
interviewees, to be able to find the right AI tool and get it to work but also to keep this AI tool
maintained. The perception of the utilization of AI was not limited to R&D and product development,
it is believed that AI can be used beyond this as well, like in post-marketing activities to understand
how well the product is received by the market but also information about the market share.

Despite that there might be some challenges and bottlenecks in implementing AI applications in
organizations for the stimulation of innovation, it is clear that the understanding from the sample
population interviewed is that many benefits might come from such an implementation. Innovation
can be stimulated indirectly by enabling resources and time for humans to focus on both other
leadership elements and skills to be able to increase innovation in the department.

5.2. Limitations
As with any other studies, out study is subject to a number of limitations. Our findings in the primary
data were based on relatively small samples (9 interviewees in total) and the time for data gathering
from both primary and secondary data was limited. We witnessed data saturation and information
convergence as we proceeded with the interviews with different informants and captured less and
less new ideas. But we are aware that the more informants we have, the higher level of data
saturation and higher validity we would have had. Although we strived to avoid informants’ biases
during the interviews by our interview design and execution, it is inevitable that we have
interpretative biases as authors, and this is an innate limitation of using grounded theory (Randall &
Mello, 2012). Lastly, as an exploratory case study based on a single company, we recognize the
limitation for generalizability. However, the aim of this exploratory case has been achieved through
giving new insights, while further research can involve in identification of empirical regularities in
other sectors or theory testing.

5.3. Conclusion and Future Research

This study is focused on the exploration and investigation of how AI applications impact the
management and leadership, and how AI can be utilized to stimulate innovation in R&D department.
The selected target company Thermo Fisher Scientific is positioned in the life science sector which
has strong linkages with AI and has tremendous transformational potential, for example the
aforementioned drug manufacturing. Under the mega-trend of the AI implementation in different
business functions, especially the innovation-driven functional unit R&D, we are hopeful to bring
novel and valuable insights of the interactions between AI and R&D management and leadership, as
well as the inputs of how AI can be taken advantage of in stimulating innovation in R&D.

For future researchers, we have attempted to explain the phenomenon of increasing application of
AI and shift in management and leadership with the results and analyses from our exploratory case

47
study focusing specifically in R&D department. The aggregated dimensions and the proposed
dynamic model can hopefully be referred to for further theoretical investigation, such as verification
and validation on the moderation and mediation effects of different factors, either partially or
holistically. AI ethics, which has not been addressed specifically in this case study, can also be further
investigated as AI ethics is evolving with different dynamics from governments, businesses, and other
organizations since it is still one thing that would affect decision making and behaviour of business
leaders.

For practitioners, this study has provided contributions in multiple aspects. Firstly, this study gives
fundamental understanding of current AI capacities and limitations to those who do not have a clear
picture, in order for practitioners to have a realistic yet possibly ambitious goal setting when
proceeding with AI investment and implementation in their own organizations on the way forward.
Secondly, our study sheds light on the AI impacts against R&D management and leadership and this
can help R&D leaders, but not limited to only R&D leaders, better understand what kinds of benefits
and challenges in people, process, technology and strategy they may face during AI implementation.
The identified shift in leadership elements will also be valuable for R&D leaders to understand the
trend and prepare for future changes by focusing more on their soft skills such as humility and
empathy. Moreover, the dynamic model can support in justifying the AI application implementation
in R&D and facilitate the change management which is necessary for people’s acceptance and
engagement. Last but not least, there are controversies how a case study with idiosyncratic findings
can be generalized to other domains (Hagg & Hedlund, 1978; Hillebrand, et al. 2001). However, it is
also argued that a lot of concepts and processes are lookalike or even structurally equivalent across
domains (Morgeson & Hofmann, 1999; Gioia, et al. 2012) and can even be more generalizable than
quantitative studies based on big survey data with statistical analyses (Tsang, 2014). In light of R&D
processes in life science sector, they are in much similar structures or forms, if not equivalent. It is
therefore senseful to consider there is transferability of the findings in this study to other companies
in the life science sector.

48
References
Adelman, C., Jenkins, D. & Kemmis, S., 1976. Rethinking case study: notes from the second
Cambridge conference.. Cambridge Journal of Education.

Anon., 2019. AGROSAVIA boosts the agricultural sector by creating efficient crops with
SampleManager LIMS software in Colombia, s.l.: Thermo Fisher Scientific.

Anon., 2021. Investor Overview. [Online]

Available at: https://ir.thermofisher.com/investors/Investor-Overview/default.aspx

Anyoha, R., 2017. The History of Artificial Intelligence. [Online]

Available at: http://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/
[Accessed 11 2 2021].

Bass, B., 1985. Leadership and performance beyond expectations. New York: Free Press.

Besanko, D., Dranove, D., Shanley, M. & Schaefer, S., 2016. Economics of Strategy. 7 ed. Versailles:
Wiley.

BioSpace, 2019. Thermo Fisher Scientific Unveils Metrios AX - a Machine Learning S/TEM. [Online]
Available at: https://www.biospace.com/article/releases/thermo-fisher-scientific-unveils-metrios-ax-a-
machine-learning-s-tem/
[Accessed 30 4 2021].

Blackburn, M., Alexander, J., Legan, J. & Klabjan, D., 2017. Big Data and the Future of R&D
Management - The rise of big data and big data analytics will have significant implications. Research-
Technology Management, 60(5), pp. 43-51.

Bonoma, T. V., 1985. Case Research in Marketing: Opportunities, Problems, and a Process. Journal of
Marketing Research, 22(2), pp. 204-206.

Brock, J. K.-U. & von Wangenheim, F., 2019. Demystifying AI: What Digital Transformation Leaders
Can Teach You about Realistic Artificial Intelligence. California management review, 61(4), pp. 110-
134.

Burns, J. M., 1978. Leadership. 1st ed. New York: Harper & Row.

Buttles-Valdez, P., Svolou, A. & Valdez, F., 2008. A Holistic Approach to Process Improvement
Using the People CMM and the CMMI-DEV: Technology, Process, People, and Culture, the Holistic
Quadripartite. Software Engineering Institute, 3.

Cardoso, D. & Ferreira, L., 2021. Application of Predictive Maintenance Concepts Using. Applied
Sciences, 11(18), p. 18.

Chamorro-Premuzic, T., Wade, M. & Jordan , J., 2018. As AI Makes More Decisions, the Nature of
Leadership Will Change. [Online]
Available at: https://hbr.org/2018/01/as-ai-makes-more-decisions-the-nature-of-leadership-will-change

Chandler, A., 1962. Strategy and Structure: Chapters in the History of the American Industrial
Enterprise. Cambridge: MIT Press.

49
Charmaz, K., 2014. Constructing Grounded Theory. s.l.:SAGE.

CIO, 2020. How empathy drives IT innovation at Thermo Fisher Scientific - CIO Ryan Snyder shares
his insights on IT’s expanding role in product development and the future of the CIO role.. [Online]
Available at: https://www.cio.com/article/3596513/how-empathy-drives-it-innovation-at-thermo-
fisher-scientific.html
[Accessed 30 4 2021].

Cooper, W., Leavitt, H. & Shelly, M., 1964. New Perspectives in Organization Reseach. New York:
John Wiley.

Cresswell, J. & Plano Clark, V., 2011. Designing and conducting mixed method research. 2nd ed.
Thousand Oaks: CA: Sage.

Cromb, D., 2007. Empowerment and Innovation: Managers, Principles and Reflective Practice.
Leadership & Organization Development Journal, 28(1), pp. 98-99.

David J. Hughes, A. L. A. W. T. A. N. A. L., 2018. Leadership, creativity, and innovation: A critical

review and practical recommendations. The Leadership Quarterly, 29(5), p. 549–569.

Dickson, B., 2019. TechTalks. [Online]

Available at: https://bdtechtalks.com/2019/11/18/what-is-symbolic-artificial-intelligence/

Draucker, C. B., Martsolf, D. S., Ross, R. & Rusk, T. B., 2007. Theoretical Sampling and Category
Development in Grounded T. Qualitative Health Research, pp. 1137-1148.

DuBrin, A. J., 2013. Principles of Leadership. 7th ed. South-Western: Cengage Learning.

Economist Intelligence Unit, 2012. The Deciding Factor: Big data and decision-making, s.l.:
Capgemini Consulting.

Editorial, 1992. Artificial Stupidity. The Economist, Volume 7770, p. 14.

Farrington, T. & Alizadeh, A., 2017. On the Impact of Digitalization on R&D - R&D practitioners
reflect on the range and type of digitalization’s likely effects on R&D management. Research-
Technology Management, 60(5), pp. 24-30.

Favaro, K., Rangan, K. & Hirsh, E., 2012. Strategy: An Executive's Definition. Strategy + Business, p.
67.

Fiedler, F. & Garcia, J., 1987. New Approaches to Effective Leadership: Cognitive Resources and
Organizational Performance. New York: Wiley.

Flick, U., 1992. Journal for the Theory of Social Behavior, Volume 22, pp. 175-98.

Floridi, L., 2014. The 4th Revolution. 1 ed. Oxford: Oxford University Press.

Foley, G. & Timonen, V., 2015. Using Grounded Theory Method to Capture and Analyze Health Care
Experiences. ealth services research, p. 1195–1210.

Freedman, D., 2019. Nature. [Online]

Available at: https://www.nature.com/articles/d41586-019-03846-0

50
Ghauri, P., Grønhaug, K. & Strange, R., 2020. Research Methods in Business Studies. 5 ed.
s.l.:Cambridge University Press.

Gioia, D. A., Corley, K. G. & Hamilton, A. L., 2012. Seeking Qualitative Rigor in Inductive Research:
Notes on the Gioia Methodology. Organizational research methods, 16(1), pp. 15-31.

Gioia, D. A. & Pitre, E., 1990. Multiparadigm Perspectives on Theory Building. The Academy of
Management review, 15(4), pp. 584-602.

Glaser, B. G. & Strauss, A., 1967. The discovery grounded theory: strategies for qualitative inquiry.
Chicago: Aldin.

Gritzo, L., Fusfeld, A. & Carpenter, D., 2017. Success Factors in R&D Leadership: Leadership Skills
and Attributes for R&D ManagersAnalysis of data from a large-scale survey reveal the behaviors,
skills, and attributes that distinguish successful R&D leaders. Research technology management,
60(4), pp. 43-52.

Guba, E. G. & Lincoln, Y. S., 1981. Effective evaluation: Improving the usefulness of evaluation
results through responsive and naturalistic approaches.. The Jossey-Bass higher and adult education
series and the Jossey-Bass social and behavioral science series.

Haenlein, M. & Kaplan, A., 2019. A Brief History of Artificial Intelligence: On the Past, Present, and
Future of Artificial Intelligence. California management review, 61(4), pp. 5-14.

Hagendorff, T., 2020. The Ethics of AI Ethics: An Evaluation of Guidelines. Minds and machines
(Dordrecht), 30(1), pp. 99-120.

Hagg, I. & Hedlund, G., 1978. Case Studies in Accounting Research. Accounting, Organization and
Society, 4(1), pp. 135-143.

Hay Group, 2005. Hay Groups Eighth Annual Analysis Of Most Admired Companies In FORTUNE
Magazine Ranking. [Online]
Available at: https://www.hrhub.com/doc/hay-groups-eighth-annual-analysis-of-most-adm-0001
[Accessed 8 March 2005].

Hillebrand, B., Kok, R. & Biemans, W., 2001. Theory testing using case studies: A comment on
Johnston, Leach and Liu. Industrial Marketing Management, Volume 30, pp. 651-657.

Hirschman, E., 1986. Humanistic inquiry in marketing research: philosophy, method, and criteria.
Journal of Marketing Research, Volume 23, pp. 237-249.

Host, A., 2021. Amount of information globally 2010-2024. [Online]

Available at: https://www.statista.com/statistics/871513/worldwide-data-
created/#:~:text=Amount%20of%20information%20globally%202010%2D2024&text=The%20total%
20amount%20of%20data,ever%2Dgrowing%20global%20data%20sphere.
[Accessed 11 2 2021].

Huang, M.-H., Rust, R. & Maksimovic, V., 2019. The Feeling Economy: Managing in the Next
Generation of Artificial Intelligence (AI). California management review, 61(4), pp. 43-65.

Hufschmitt, E.-M., 2019. AI is the megatrend that will shape the future. [Online]
Available at: https://www.abbyy.com/blog/ai-is-the-megatrend-that-will-shape-the-future/

51
IBM , 2020. What is strong AI?. [Online]
Available at: https://www.ibm.com/cloud/learn/strong-ai

Investopedia, 2021. Investopedia. [Online]

Available at: https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp
[Accessed 2021].

Joseph, B. et al., 2015. Digital Vortex - How Digital Disruption Is Redefining Industries. Global
Center For Digital Business Transformation, 6.

Judge, T. A., Bono, J. E., Ilies, R. & Gerhardt, M. W., 2002. Personality and Leadership: A
Qualitative and Quantitative Review. Journal of applied psychology, 87(4), pp. 765-780.

Judge, T. A., Colbert, A. E. & Ilies, R., 2004. Intelligence and Leadership: A Quantitative Review and
Test of Theoretical Propositions. Journal of applied psychology, 89(3), pp. 542-552.

Keller, R. T., 1992. Transformational Leadership and the Performance of Research and Development
Project Groups. Journal of Management, 18(3), pp. 489-501.

Kleefstra, A., 2019. A Literature Review into Leadership Styles Discussed in the Past Five Years.
Open Journal of Social Sciences, Issue 7, pp. 180-190.

Kruse, K., 2013. What Is Leadership?, s.l.: Forbes.

LeCompte, M. & Goetz, J., 1982. Problems of reliability and validity in ethnographic research. Review
of Educational Research, 52(1), pp. 30-60.

Lincoln, Y. & Guba, E., 1985. Naturalistic Inquiry. Newbury Park: Sage.

Liu, J. C. H. F. J. Y.-L. Y. B., 2020. Influence of artificial intelligence on technological innovation:

Evidence from the panel data of china's manufacturing sectors. Technological forecasting & social
change, Volume 158, pp. 120-142.

Maltaverne, B., 2015. People, process and technology: The 3 crucial ingredients for a successful
change management. Pool4Tool Procurement, 14 9.

Mansfield, E., 1998. The Speed and Cost of Industrial Innovation in Japan and the United States:
External vs. Internal Technology. Management Science, 34(10), pp. 1157-1168.

Marshall, C. & Rossman, G., 1989. Designing Qualitative Research. Newbury Park: Sage.

McNamara, A., Smith, J. & Murphy-Hill, E., 2018. Does ACM’s Code of Ethics Change Ethical
Decision Making in Software Development?. New York, ACM.

Merriam, S. B., 1998. Case Studies as Qualitative Research. In: Quantative research and case study
applications in education. San Francisco: Jossey-Bass Inc, pp. 26-42.

Metcalf, L., Askay, D. A. & Rosenberg, L. B., 2019. Keeping Humans in the Loop: Pooling
Knowledge through Artificial Swarm Intelligence to Improve Business Decision Making. California
management review, 61(4), pp. 84-109.

Miailhe, N. & Hodes, C., 2017. The Third Age of Artificial Intelligence. Field actions science reports,
pp. 6-11.

52
Miles, M. & Huberman, M., 1994. Qualitative Data Analysis - An Expanded Sourcebook. 2 ed.
Thousand Oaks: Sage.

Mitchell, V.-W., 1993. Industrial in-depth interviews: some considerations for first-time users.
Marketing Intelligence & Planning, pp. 25-29.

Morgeson, F. P. & Hofmann, D. A., 1999. The Structure and Function of Collective Constructs:
Implications for Multilevel Research and Theory Development. The Academy of Management review,
24(2), pp. 249-265.

Morse, J. M., 1994. Designing funded qualitative research. Handbook of qualitative research , p. 220–
235.

Nag, R., Corley, K. G. & Gioia, D. A., 2007. The intersection of organizational identity, knowledge,
and practice: Attempting strategic change via knowledge grafting. Academy of Management Journal,
Volume 50, pp. 821-847.

Norris, W. R. & Vecchio, R. P., 1992. Situational Leadership Theory: A Replication. Group &
Organization Management, 17(3), pp. 331-342.

Odumeru, J. A. & Ogbonna, I. G., 2013. Transformational vs. Transactional Leadership Theories:
Evidence in Literature. International Review of Management and Business Research, 2(2).

Office of Goverment Commerce, 2007. The Introduction to the ITIL Service Lifecycle. s.l.:The
Stationery Office.

Palinkas, L. A. et al., 2015. Purposeful Sampling for Qualitative Data Collection and Analysis in
Mixed Method Implementation Research. Administration and policy in mental health and mental
health services research, 42(5), pp. 533-544.

Patton, M., 2014. Purposeful sampling and case selections: Overview of strategies and options.
Qualitative Reserach & Evaluation Methods, pp. 264-272.

PEARCE, C. L., WASSENAAR, C. L. & MANZ, C. C., 2014. IS SHARED LEADERSHIP THE
KEY TO RESPONSIBLE LEADERSHIP?. The Academy of Management Perspectives, 28(3), pp.
275-288.

Pearl, J. & Mackenzie, D., 2018. The Book of Why - The New Science of Cause and Effect. 1 ed.
s.l.:Penguin Random House UK.

Pearson, J., 2009. People-Process-Technology - The eternal triangle. [Online]

Available at: https://deconstructingitsm.wordpress.com/2009/02/17/people-%E2%80%93-process-
%E2%80%93-technology-the-eternal-triangle/
[Accessed 13 3 2021].

Peräkylä, A., 1977. Reliability and validity in research based on tapes and transcripts. Qualitative
Research: Theory, Method and Practice.

Pieterse, A. N., Knippenberg, D. V., Schippers, M. L. & Stam, D., 2010. Transformational and
transactional leadership and innovative behavior: The moderating role of psychological empowerment.
Journal of Organizational Behavior, Volume 31, p. 609–623.

Pisano, G. P., 2012. Creating an R&D Strategy. s.l.:Harvard Business School.

53
Porter, M., 1996. What Is Strategy. Harvard Business Review, pp. 61-78.

Prentice, W., 2004. Understanding Leadership. Harward Business Review.

Prodan, M., Prodan, A. & Purcarea, A. A., 2015. Three new dimensions to People, Process,
Technology improvement model. Advances in Intelligent Systems and Computing, 1.

Randall, W. S. & Mello, J. E., 2012. Grounded theory: an inductive method for supply chain research.
International Journal of Physical Distribution & Logistics Management, 42(8/9), pp. 863-880.

Rangaiah, M., 2020. How Tesla is making use of Artificial Intelligence in its operations?. [Online]
Available at: https://www.analyticssteps.com/blogs/how-tesla-making-use-artificial-intelligence-its-
operations
[Accessed 25 February 2020].

Rijanto, A., 2018. Innovation Driven Enterprise, Sustainable Business and Firm Financial
Performance. The Asian Journal of Technology Management, 11(1), pp. 10-25.

Robson, C., 1993. Real World Research. A Resource for Social Scientists and Practitioner
Researchers.. Oxford: Blackwell Publishers Inc..

Scheidlinger, S., 1994. The Lewin, Lippitt and White study of leadership and “social climates”
revisited. International journal of group psychotherapy, pp. 123-127.

Schmidt, F. L. & Hunter, J. E., 1998. The Validity and Utility of Selection Methods in Personnel
Psychology: Practical and Theoretical Implications of 85 Years of Research Findings. Psychological
bulletin, 124(2), pp. 262-274.

Schuklenk, U., 2020. On the ethics of AI ethics. Bioethics, 34(2), pp. 146-147.

Shrestha, A., 2020. Machine Learning: A Primer to Laboratory Applications. [Online]

Available at: https://www.thermofisher.com/blog/connectedlab/machine-learning-a-primer-to-
laboratory-applications/
[Accessed 22 01 2020].

Shrestha, Y. R., Ben-Menahem, S. M. & von Krogh, G., 2019. Organizational Decision-Making
Structures in the Age of Artificial Intelligence. California management review, 8, 61(4), pp. 66-83.

Singh, A., 2019. Prescouter. [Online]

Available at: https://www.prescouter.com/2019/06/pharmaceutical-rd-global-spending-trends-in-2019/

Sjödin, D. R., Parida, V., Leksell, M. & Petrovic, A., 2018. Smart Factory Implementation and
Process Innovation. Research-Technology Management, 61(5), pp. 22-31.

Sterne, J., 2017. Artificial Intelligence for Marketing. 1 ed. s.l.:John Wiley & Sons.

Strauss, A. & Corbin, J., 1990. Basics of Qualitative Research: Grounded Theory Procedures and
Techniques. s.l.:SAGE Publications.

Strauss, A. L. & Corbin, J. M., 1998. Basics of Qualitative Research: techniques and procedures for
developing grounded theory. 2nd ed. Thousand Oaks: SAGE.

54
Tambe, P., Cappelli, P. & Yakubovich, V., 2019. Artificial Intelligence in Human Resources
Management: Challenges and a Path Forward. California management review, 61(4), pp. 15-42.

Thaddeus Hunt, L. F., 2018. Leadership: Past, Present, and Future: An Evolution of an Idea. Journal
of Arts and Humanities, Volume 08(Issue 02).

Thermo Fisher Scientific , 2019. Artificial intelligence tools for Amira-Avizo Software and PerGeos
Software. [Online]
Available at: https://www.thermofisher.com/se/en/home/industrial/electron-microscopy/electron-
microscopy-instruments-workflow-solutions/3d-visualization-analysis-software/amira-avizo-pergeos-
artificial-intelligence.html
[Accessed 30 4 2021].

Thermo Fisher Scientific, 2020. The Power of AI in Drug Discovery and Development. [Online]
Available at: https://www.thermofisher.com/blog/connectedlab/the-power-of-ai-in-drug-discovery-
and-development/
[Accessed 24 4 2021].

Thermo Fisher Scientific, 2020. Using AI to Boost Crop Yields. [Online]

Available at: https://www.thermofisher.com/blog/connectedlab/using-ai-to-boost-crop-yields/
[Accessed 24 4 2021].

Thermo Fisher Scientific, 2021. Thermo Fisher Scientific Annual Report 2020. [Online]
Available at: https://s2.q4cdn.com/210843478/files/doc_financials/2020/ar/2021-Annual-Report-10k-
Bookmarked.pdf
[Accessed 22 4 2021].

Thermo Fisher Scientific, 2021. Thermo Scientific Helios 5 EXL Wafer DualBeam Offers Automation
in TEM Sample Prep to Help Accelerate Time-to-Yield - Expedite TEM preparation with machine
learning capabilities. [Online]
Available at: https://thermofisher.mediaroom.com/2021-04-20-Thermo-Scientific-Helios-5-EXL-
Wafer-DualBeam-Offers-Automation-in-TEM-Sample-Prep-to-Help-Accelerate-Time-to-Yield
[Accessed 30 4 2021].

Thomas Marshall, T. C.-L. D. C. &. D. H., 2017. Cognitive computing and eScience in health and life
science research: artificial intelligence and obesity intervention programs. Health Information Science
and Systems.

Tsang, E., 2014. Generalizing from research findings: The merits of case studies. International
Journal of Management Review, 16(4), pp. 369-383.

Turing, A., 1950. COMPUTING MACHINERY AND INTELLIGENCE. Mind, Volume 236, pp. 433-
460.

Tushman, M. L., 2017. Innovation Streams and Executive Leadership: R&D leadership plays a central
role in shaping a firm's ability to both exploit existing capabilities and explore new technological
domains. Research technology management, 60(6), pp. 42-47.

Xing, F. et al., 2020. Driving Innovation with the Application of Industrial AI in the R&D Domain.
Springer International Publishing, pp. 244-255.

Yin, R., 1994. Case Study Research - Design and Methods. Newbury Park: Sage.

55
Yin, R. K., 2009. Case Study Research: Design and Methods. In: 4th, ed. Case Study Research.
s.l.:s.n.

Yukl, G., 1989. Managerial Leadership: A Review of Theory and Research. Journal of Management,
15(2), pp. 251-289.

Zaltman, G., Pinson, C. & Angelmar, R., 1977. Metatheory and Consumer Research. New York:
Rinehart & Winston.

56
Appendixes

Appendix 1: Interview Questions

Questions related to “People” and “inform and enable”:
1. What kind of information can people in R&D get from AI?
2. Where does this information come from?
3. How can this information be processed before being transferred to people?
4. Who/which roles will use this information?
Questions related to “People” and “transform”:
5. What new knowledge and skills are people required to have in using AI?
6. What changes are needed in the planning of employee development to use AI?
7. What changes are needed in talent acquisition?
8. How do you think people's motivation is affected when working with AI?
9. How does AI change the knowledge sharing?
Questions related to “Process” and “inform and enable”:
10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being
applied?
11. How does AI support these processes?
12. What are the benefits?
Questions related to “Process” and “transform”:
13. How will AI change the way R&D work is being performed?
14. How will AI change the tools that R&D employees are utilizing?
Questions related to “Technology” and “inform and enable”:
15. Does AI provide information for understanding and predicting new technology trends? How?
Questions related to “Technology” and “transform”:
16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?
Questions related to “Strategy” and “inform and enable”:
The strategy of Thermo Fisher Scientific:
17. How can AI contribute to customer-centric innovation?
18. How can AI contribute to the high growth markets and to the emerging markets?
19. How can AI contribute to delivering unique value proposition to customers?
Questions related to “Strategy” and “transform”:
20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?
21. How can AI applications in R&D influence the revenue?
22. How can AI influence the R&D investment?
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)

Questions related to leadership and innovation:

24. How much do you involve your employees in decision making?
25. What do you think motivates your employees?
26. How do you react to mistakes/weaknesses of your employees?
27. Are you more oriented in the end result or the process in itself?
28. How do you see AI implementation affects your leadership?
29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?
30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI?

57
Appendix 2: Interview Notes
Interview date: 12/4/2021

Interviewee 1:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:

1. What kind of information can people in R&D get from AI?

It depends on the product, it could be about users of the products or markets, can be also from
operators.

It could be figures or text for people to respond faster and safer.

2. Where does this information come from?

When it comes to product, the information can be from customers using the product, like customer
behavior pattern.
3. How can this information be processed before being transferred to people?

need to have systems similar to data management to read and categorize information

4. Who/which roles will use this information?

- Someone who wants to be creative and design product, entrepreneurial minded to make
product faster and safer and quicker to market
- Roles with hands on skills, e.g. product designers

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?

- Need to have analytical skills

- Understanding of reliability of the factors and risks
- Judging if the source is good. What can be done with the AI.

6. What changes are needed in the planning of employee development to use AI?
- Analytical skills are needed
- Leaders need to understand what skills are needed according to the changes and how to
make the changes. It is not just about AI but also for other new challenges.
- Leaders need to think about how to hire the employees with the required skills. We cannot
only have people coming from fresh graduation.
- It is more about how to sell the use of AI to employees internally, convincing them that this is
not going to take their work away, but to make sure they are open to this new thing that can
make their work more efficient (entrepreneurial skills and creative skills, learnability)

7. What changes are needed in talent acquisition?

58
 To make people easy to be on board and give people education on how the system works and how
to use the system.

we cannot only hire people who are newly graduated or have been working with this for a few years,
it is also about hiring people who are adaptive to changes and different way of learning.
8. How do you think people's motivation is affected when working with AI?

It is important to make the employees informed why and how it is going to be implemented, on
how it is going to help customers, colleagues and processes.

It is almost crucial with change management and to state that this is not going to take away their
job, or someone is going to monitor your whereabouts.

It can go both way in terms of motivation or demotivation. For people who are open minded it
would be motivating, but for people who consider this as a threat it would be demotivating.

That is why we need all people in the organization to start touching on this and start the discussion
9. How does AI change the knowledge sharing?

AI can provide stronger evidence when people want to argue for their points, e.g. having some
functions or doing things in a new way

Data is existing but it does not tell consumer behaviors like how people are using our products. But
AI can give stronger points

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?
(unasked)

11. How does AI support these processes? (unasked)

12. What are the benefits? (unasked)

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?

Providing R&D extra layers of information, or approval or evidence to confirm their decisions that
this is the right way to go in product design with enough information about customer product usage,
or how operators and engineers do with the products.

14. How will AI change the tools that R&D employees are utilizing?

Not sure but tools for sure will be changed.

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?

59
 - I am sure it will and that is how we can benefit from it. AI will create next user level and next
trend

Questions related to “Technology” and “transform”:

16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?

- I will it will change absolutely.

Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer centric innovation?

- AI provides information about how customers are using our products, how to provide them
with faster answers, or how to train customers with manuals, and being more proactive in
providing safety

18. How can AI contribute to the high growth markets and to the emerging markets?

High growth markets may be more biased and conservative than emerging markets which are more
open

We need more time led to work with what they have

19. How can AI contribute to delivering unique value proposition to customers?

AI can make products more personalized as we have more information about what their need is and
their usage

the product can be the same with everyone’s but we can understand the customers better so that
we can focus more on their pain points or the features that are really beneficial for them

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?

- It depends on how the organization is picking up on the AI.

- “In or not” is really the decision for the company to follow, ‘’whether AI should rule the
decision making
- But normally decisions are having stronger ground and more informative.
- AI can give faster and more reliable and more argumentative way of decisions, but education
is essential to understand how those decisions are made.

21. How can AI applications in R&D influence the revenue?

AI is positive in influencing revenue with its providing of faster decisions, adapting faster to
customers and products in the markets, making it possible to shift before you have seen the real shift
in real life, especially for large organizations which normally consume more time to get things

60
through
22. How can AI influence the R&D investment?

First investment needs to be in place for AI applications.

But AI can get people more confirmed in which way the investment should be made, maybe not
only relying on traditional KPIs, and having the information more in the face of investors or the
financial powers of the organizations, and seeing the shift earlier
23. Does AI influence organization structure? How? (More integration? centralized or de centralized)
The needed skills might be different but the people’s roles would not be changed automatically

The number of the positions should maintain the same, but people might change their job or ways
to do work. E.g. business analyst can be having the same role but the way of working can be changed
with AI applications, having faster and more secure and accurate information.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?

The team will figure out how to do their tasks; As long as the results are delivered, the “how” part is
decided by the employees themselves

People in the team will also be involved in discussing e.g. the structuring of the team and new
recruitment where challenges and strengths, as well as positioning of people will be discussed.

Wherever possible, team will be involved.

However, as big global organization, many tasks are set according to established processes and
procedures, unless it is something operational and every day working life.

25. What do you think motivates your employees?

When people are working with their specific interested areas that drive them, they will be
motivated. E.g. engineers like to perform troubleshooting and not administrating, commercial people
like to work with targets such as financial responsibility and having key accounts.

The team has different specialties in each member.

26. How do you react to mistakes/weaknesses of your employees?

That is part of human being and how we learn. We, not only as individual but also as organization,
can learn from mistakes and focus on what needs to be done to avoid those mistakes. Progress could
not be made if we stand still.
27. Are you more oriented in the end result or the process in itself?

End results. How to get things done depends on the persons themselves.
28. How do you see AI implementation affects your leadership?

I see that is an absolute coming trend. AI can assist in product in terms of speed and safety.

61
 AI is not reaching to where I am so far, and my perception that it is currently kept in R&D and other
parts of the organization.

But I think we should start addressing it sooner and getting familiar with it to understand how it can
help us, which skills we need to have since we will have big impact from it.

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?

We need to have support, investment, and people who are open and willing to try and make the
efforts.

change management is needed through the different organizations.

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

Interview date: 13/4/2021

Interviewee 2:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:

1. What kind of information can people in R&D get from AI?

In R&D, physical product that is going to be used by human (instrument, car etc.) needs human
interactions. There should be important developments in safety, ergonomics, making it simple.

AI is helping speed up the process of how humans are interacting with products. Analysis to be done
in the new product from the customer behaviors.

Example: Software capturing eye movements from people, or how many clicks in a page to
understand their behavior
2. Where does this information come from?

Software, or systems capturing or collecting data of users using the products to improve safety.

Example: physical product using software and camera to capture behavioral patterns.
3. How can this information be processed before being transferred to people?

Example: bioinformatics data of trillions of data points is put together in an interpretable way. There
should be a clever way of seeing the patterns in AI.

From our company we are using statistics from our users on what they purchase and look out in our
workshop. With this data the analysis will be used in campaigns so that we can lead them into
defining other products of the similar areas. One question leads to another question in a chain.
“Pattern” should be captured from AI.

62
 Aim of the data should be defined first for AI to process, e.g. to make the product with certain
features (light weight, user friendliness)
4. Who/which roles will use this information?

Both managers and employees definitely will use this information.

For management, information regarding management decisions is interesting to make more

customers buying the products.

On the other hand, to the basic level information is of high interest.

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?

Hopefully no new skills will be needed. Human does not have the capacity of processing the large
data. AI can do something that we cannot do with their exceptional capacity to deal with big data
and come up with suggestions (e.g. machine for disease diagnosis). We can stay where we are.

People can see this as skeptical since it seems a threat. Training will probably be needed to avoid
people to jump to conclusion to see AI as replacement of their job.
6. What changes are needed in the planning of employee development to use AI? (unasked)
7. What changes are needed in talent acquisition? (unasked)
8. How do you think people's motivation is affected when working with AI?

Depends on how it is reported in general media on AI and how the communication is conveyed.
There will be a mix of positivity and negativity.

Elderly people do not have smartphone; we need to give alternative for these people to do things.
9. How does AI change the knowledge sharing? (unasked)

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?
(unasked)
11. How does AI support these processes? (unasked)
12. What are the benefits? (unasked)

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?

The aim of using AI in R&D is to develop products faster, less cost (e.g. labor cost) in the long run.

With using AI, maybe there can be a shift that product produced in high labor cost countries could
also be produced in cheap labor countries. A geographic shift of R&D activities for the company is
possible.
14. How will AI change the tools that R&D employees are utilizing? (unasked)

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?

63
That is a challenge. I am skeptical on AI’s ability to imagine or taking leaps and new directions.

Questions related to “Technology” and “transform”:

16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?

It is a wide question. Example: pharmaceutical companies are doing tests on animals and now with
AI they can do virtual experimentation and simulation to reduce the number of tests done on
animals.

Maybe this can bring changes in investing new technology in producing new products. Physical lab
can be replaced by AI to reduce operational costs and labor costs.

Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer centric innovation?

AI can help analyze what customers are buying and also information shared by customers using our
webpages; and give good suggestions to customers to buy products to help them in their workflows
18. How can AI contribute to the high growth markets and to the emerging markets? (unasked)
19. How can AI contribute to delivering unique value proposition to customers?

By giving good product information and suggestions to customer by using AI, customers can get
improvements such as faster process and get more effective and efficient results.

It can also be a challenge for AI to tap into customers’ pain points, like being stuck in analysis or
seeking new ways to do analysis. Information seeking in this area to understand customer’s pain
points is valuable and good to be developed in the future.

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?

It is a daily job for R&D to use excel and powerBI to analyze the product price, marginal costs,
market prices, and product placements to find ways to make our products competitive but not with
too low margin.

There are many people working as data controlling in prices and where to produce, profitability,
operations locations, supplementary products.

There is a lot of data in a company. The trick is to use AI to gather these data in a big data way and
do a clever analysis and build a report or tool in management decision on prices, locations, and when
to produce.
21. How can AI applications in R&D influence the revenue?

AI can get information about pricing, quality.

64
 Revenue is not the only focus; cost is also a focal point. It should be the profit that AI can contribute.
22. How can AI influence the R&D investment? (unasked)
23. Does AI influence organization structure? How? (More integration? centralized or de centralized)
(unasked)

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?

A high proportion of decisions are dependent on the team members, including information,
solutions, and background.

I am seldom working on isolation and working in a project-based environment and it is mostly based
on joint decisions.

25. What do you think motivates your employees?

Employees should have a feeling of doing something worthwhile, making a difference. Purpose is
given for the employees (to customers, society, market) to deliver good products and services.
26. How do you react to mistakes/weaknesses of your employees?

Actively create an environment where people feel safe for mistaking, learn by the mistakes, make
employee free of fear to commit mistakes, not be punished but delivering training instead.

I try to be humble and admitting mistakes, and set expectation and not lower the standard. A
balance should be found there.
27. Are you more oriented in the end result or the process in itself? (unasked)
28. How do you see AI implementation affects your leadership?

AI technology is coming up and moving into useful tools in different areas. It is definitely to be used
in the future to speed up processes in the future which are currently done manually today.

It can lead to selection of answers to the questions, building up to decisions.

It is definitely a tool to provide information, faster management

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI? (unasked)

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

Interview date: 15/4/2021

Interviewee 3:

Interview Questions related to AI applications vs R&D management activities:

65
Questions related to “People” and “inform and enable”:
1. What kind of information can people in R&D get from AI?

-There should be plenty of information

- Footprint; the benefits of how it can be used; software that can be helpful: e.g. 2D heat maps of
protein analysis.
2. Where does this information come from?

-It depends on what we are looking for, it can come from anywhere.

- From Customer (from survey), from end users.

- Critical mass is also very beneficial to get information.

3. How can this information be processed before being transferred to people?

- It might be difficult. We have to think about the streamlining very much, and what we are
interested in the end. First for AI, it can be about possibility, feasibility and what it takes to do it, and
whether it could help us in the end.

- There are thousands of pieces of information for R&D to receive and AI can help in funneling down
the options and give people needed suggestions.

4. Who/which roles will use this information?

-Definitely the managers, not just R&D. it will also be Product Management and Marketing.

- All the information should be brought into a committee composed of different departments’
management such as R&D, Product Management and Marketing. Information can come first from
Marketing, to PM and then to R&D for why we should do this work/develop the new product.

- Manufacturing is also a critical part, with AI’s help.

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?

-Technically, how it can be used needs to be learned about.

- Data analysis needs to be trained as well

- It would be technically challenging to get all the skills and knowledge to be trained to people but it
would not be difficult to understand it. AI’s purpose is to make our lives easier.
6. What changes are needed in the planning of employee development to use AI? (unasked)
7. What changes are needed in talent acquisition?

- Best success is to include people working in different ways with different ideas.

- But not forgetting people having experience.

66
- When people with new ideas/backgrounds and people with experience meet, we can make
something work. E.g. in marketing department we have a new person previously from traveling
agency who is good at digital marketing.
8. How do you think people's motivation is affected when working with AI?

- AI is very new and exciting and interesting to people. And people would like to learn more, e.g. AI in
Marketing.

- It is a trend, and there has been a lot of progress.

9. How does AI change the knowledge sharing?

- With all the data shared inside the organization, we are much quicker in finding solutions.

- It has changed the way in looking forward together for new possibilities, new improvement ideas.

- it will save time in knowledge sharing between different groups.

- Non-limiting (24/7) access can help knowledge sharing in different geographies.

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?

-data collection part to justify the project

-All the milestones in the project are critical in using AI (early development process, validation
process, manufacturing, quality assurance).
11. How does AI support these processes? (unasked)
12. What are the benefits? (unasked)

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?

-it will be much quicker and more efficient.

- it will be easier to make decisions to move forward.

- it will be helpful in detecting errors and monitoring processes

14. How will AI change the tools that R&D employees are utilizing?

- monitoring processes, simplifying decision making processes, quality control and where errors
occur.

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?

-Absolutely. AI can simply things by e.g. adding new possibilities or eliminating things that do not give
added values.

- We can streamline what we are interested in by using AI by filtering the noise.

67
Questions related to “Technology” and “transform”:
16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?

-I think R&D will see the opportunities and benefits AI brings. R&D will move forward because AI
gives opportunity for easier access and easier ways of working in the future. There will be interest in
investing in this area.

Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer-centric innovation?

-For customer, if we with AI can provide what we are selling and let customer understand how we
are benefiting their projects, that would be something we will benefit from.
18. How can AI contribute to the high growth markets and to the emerging markets?

-In high-growth markets, new things and technologies are taken in a faster pace. They are the early
adopters. They will be quickly jumping into new technologies that make things simplified and easier.

- In emerging markets things will happen slower and they will adopt them but it will take longer time.
19. How can AI contribute to delivering unique value proposition to customers?

- AI can provide new value proposition, unique selling points by showcasing them to the customers

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?

-We need to get hold all the data and look into them to convey the message to higher management.

- AI can showcase the summarized information to GLT or decision makers, which makes it easier for
managers to explain the logics behind. We do not need to show all the data points but only the
important ones which AI helps you in identifying.
21. How can AI applications in R&D influence the revenue?

- AI can interpret information for management from massive data.

- AI can help revenue increase in different positions: sales, new product development, etc. AI can
close the information loop by feeding back information from one end to another end. In this way it
can help bring more revenue to the company.
22. How can AI influence the R&D investment?

- When you need to make a new investment, it will need massive data and justify all the data which
AI can provide its help with, in an understandable way. The database is big and the significance of
the data will be high and validity is therefore high.
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)

68
- A lot of processes will be changed, and will be easier.

- Maybe certain types of job will not be relevant for the future, meaning certain roles will be replaced
by AI.

- It will not happen overnight but gradually. Companies should educate people in the areas which can
be replaced by AI to work in other areas so that they can work in somewhere different.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?

- We do that quite a lot, because we need to take all the data, for example sales and revenue, into
decision making.

- We need to be agile as well

- Involvement of team will be on daily basis.

25. What do you think motivates your employees?

- If they have done a good job, tell them and appreciate them with e.g bonus, that will motivate
them very much.

- working as a team and seeing impact to other people, e.g. vision to help the world become
healthier, safer and cleaner.

- working towards a same goal. If the team knows where the direction is (e.g. where we will be in 5
years) it would be a good motivator.
26. How do you react to mistakes/weaknesses of your employees?

- I think it would be good to talk about them straight away. Everybody makes mistakes and it should
be that the people know that it is not the one-man job but it is teamwork and individual would not
be the only responsible person when there is a mistake.

- We are all into this together, we need to be honest and try to rectify.

27. Are you more oriented in the end result or the process in itself? (unasked)
28. How do you see AI implementation affects your leadership?

- The more the better. Anything that can save time and bring more efficiency will be beneficial for
leadership.

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?

- There is absolutely a lot of potential there. That can streamline the R&D and innovation. E.g. if you
have this in the initial phase of the R&D project, we used VR to look into and try to sell bio-
manufacturing process factory. VR helps in visualizing how much room different systems in need will
be occupying before the customer sees the real product. Same thing for R&D instrument, if we need

69
to build a new system, we need to evaluate how much footprint it would occupy before you see the
real thing.

- If we have a solution in AI way, then we will know if we should move forward or not, “make it or
break it”, if we should invest more money to continue with your R&D projects.

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

Interview date: 16/4/2021

Interviewee 4:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:

1. What kind of information can people in R&D get from AI?
Data for data analysis, but also data for customer experience. Get information from customers in
another way, to be able to develop better and more spot-on products. From two ends, one is the end
customer. To analyze how they might want to use our products. The order end is when you develop a
product, it is a lot of data to consider and AI can be used to help analysis of this data.
2. Where does this information come from?
Customer
3. How can this information be processed before being transferred to people?
To make it visual, visualize it. Will have much and already have much information. Present it in a
processed, nice, easy to understand and visual way. Specially managers are exposed to a lot of
information and have few minutes to spend on understanding the datapoints.
4. Who/which roles will use this information?
Will use it on both (multiple levels), from customers to understand what to develop. Can also be used
in the end of the development phase, when having all data and to be able to digest it and present it
to top management, to authorities. This to show that you have a product that have value for
management or provide data to show that it is a safe product. For authorities to approve that a
product can be sold. The employees will be able to use this information in the beginning to
understand which kind of product we develop, and the management in the end to understand what
we will get out of the product. Managers will also be able to benefit during the process, as they visit
the project during the different phases (should prove that it has its values).

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?
Need to have skills to be able to develop and make use of the AI technology (the engineering part).
Also need to know how to interpret things, the data that is coming from AI. Engineers are needed to
understand how AI can be used to get something, but also need to train people to things coming out
of these (for managers and others that use). How do you use the data, what does it tell us?
6. What changes are needed in the planning of employee development to use AI?

70
Long plan is needed if we are going there. R&D departments are quite conservative. R&D more
conservative than you think. Need people that can drive it, but also ensure to have followers. It is a
big change for department to bring this in. But it will be very useful.
7. What changes are needed in talent acquisition?
Need to have both drivers, to get in. But also, followers, to employ followers as well.
8. How do you think people's motivation is affected when working with AI?
Will have the whole spectrum, will have those who are scared to people that will love it. This will
depend on the generation.
9. How does AI change the knowledge sharing?
AI ca be used more for collaboration. To bring work from different departments together. A
speculation is that AI can be used here to get it together and to get the bigger picture.

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?
It is easier in development then in pure research. When you have an idea and know what to do. Then
can use AI to get information. In pure research AI is of less use.
11. How does AI support these processes?
Help to collect the input requirements, help to process more. More data in beginning. Will probably
be able to reach more broader then without it. Maybe can help to keep you on track, or also close
down project that shouldn't continue.
12. What are the benefits?
Will be beneficial for those who develops things without any feelings. Can become “someone baby”.
This will influence the decision making. Other projects can compete with your “baby”. This will
influence. Bias when making decision. Will be more objective with AI. AI doesn’t have a baby.

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?
Can influence the decision process. Could help to reduce some studies, help use analyze better. Can
get carried away with what you can get from a study. Can potentially extrapolate better, can avoid
repeating studies when something has been done before.
14. How will AI change the tools that R&D employees are utilizing?
Can do more simulations in the future, don’t need to test everything. It is harder in the biological
stuff. Easier with instrument and system. More predictable from a mechanical standpoint then a
biological.

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?
Yes, it can be used to predict new technology. Maybe more then we think. It can pick up trends
quicker.

Questions related to “Technology” and “transform”:

16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?
It can predict, could be one way of predicting. We will probably still use the same methods we do to
predict but with AI emotions will not be involved.

71
Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer-centric innovation?
AI can help and evaluate, can be used to simulate. Can even simulate customers and get a bigger
user base. It is costly to go out to customer, can help getting a bigger user base.
18. How can AI contribute to the high growth markets and to the emerging markets?
Can help with the analysis of market needs, marketing analysis about the customer needs. That is
one area that it could help. Could also accelerate development and market access of a product as
well, if done wisely. It is one thing what the market needs, but also to get the access. EM are difficult
to get (our) products in to. If could get a better understanding of what we need there, it could
accelerate growth.
19. How can AI contribute to delivering unique value proposition to customers?
Could help to develop the value proposition, to make that more powerful.

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?
This will be a factor in the future, AI will support us in this decision by providing different kind of
analysis. Everything from the customer, the financial, to prediction. IT will not replace us (our brains)
but it will be another factor, another tool that we take into consideration.
21. How can AI applications in R&D influence the revenue?
Will get more out of the organization. By using good tools, with AI and other tools, and drive
efficiency.
22. How can AI influence the R&D investment?
Will take the longest to implement. Again, the analysis that will help and support us.
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)
Certain things can be influenced, as for example in a service centers (complaints and etc). Could have
AI replacing people at certain level, but the difficult things will need to be handled by people.
Routines could be delt by AI. That could influence the organization.
Data is already available but not really accessible for R&D (from marketing e.g.). AI can help here.

24. Is there anything you want to add that could be useful to the study.
Questions we could ask; how would we get this started? How do we start moving it? At some point,
we need to start moving in this direction whether we like it or not. That could be a question.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?
Always discussing things with employees, if it is decisions where employees can be involved. As a
manager, sometimes you need to take decisions where employees cannot be involved.
25. What do you think motivates your employees?
When they make progress, and sees that the products they develop comes to use. When they see
that it actually leads to a result.
26. How do you react to mistakes/weaknesses of your employees?

72
Consider mistakes as a learning point. Have an open discussion. Have weekly reviews where they go
through mistakes and learning. Believes it is important to share the mistakes and learnings with
others. Some mistakes easier to share then others.
27. Are you more oriented in the end result or the process in itself?
Personally in the Process.
28. How do you see AI implementation affects your leadership?
Have not seen it so far in my leadership but believes it will come, in the future. Will be an additional
tool for decision making in the future.

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?
The analysis part, tools that can draw conclusions quicker then we can today. Where I am working in,
it is mainly data analysis. Believe that this will differ in other areas.

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

Interview date: 22/4/2021

Interviewee 5:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:

1. What kind of information can people in R&D get from AI?

2. Where does this information come from?

Q1-Q2 Can come from instruments we use, software, technologies we have. Can also come from the
different processes.

Can come from previous studies, experiments. Someone needs to look at the data to see if it makes
sense. Someone needs to evaluate it. Validate the result and the data.
3. How can this information be processed before being transferred to people?
Have a control step here where people should validate the results. It is taken by granted that this is
being done as a part of AI. Some kind of processing in needed, to give the relevant part.

4. Who/which roles will use this information?

Both the employees (lab people) for experiments, in the leadership roles. Working with teams and
processes.

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?
Open mind. More IT skills and computer skills, to be able to take in information and interpretive the
result.

73
6. What changes are needed in the planning of employee development to use AI?
Training in how AI gathers information, so can be more observant (related to validation step) when
controlling the result. Instead of technology trainings related to the role, more training will be
needed for the AI.
7. What changes are needed in talent acquisition?
Still need competent people (experts) in the relevant field, if it is related to for ex allergy and
immunology field. However, little AI skills would be necessary (to understand the information).
Technical expertise in the relevant field. Depending on if AI can replace some specific steps of the
laboratory technicians (scientists), those skills will not be needed anymore.
8. How do you think people's motivation is affected when working with AI?
Would depend on what AI is being used for. For ex, if used for quality improvement, it would
increase the motivation. But if it is replacing people's jobs, this would have a negative effect. IT skills
would potentially affect the motivation. Not necessarily related to age, more related to interests.

9. How does AI change the knowledge sharing?

It would be good if it could increase the knowledge sharing.

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?
Both in laboratory work and in management. In project management, processes, and the
technologies in the laboratory.

11. How does AI support these processes?

New employees for management, help with onboarding. Specific learning plan for the different roles.
Could potentially automate certain steps.

12. What are the benefits?

Similar things can be applied to other processes in management or in leading positions. More
personalized decisions. Personally, I use a lot of time to figure out next steps, risk analysis and
change request. Many processes can be simplified with AI by taking decisions on the next step
instead of you. (or at least suggest decisions)

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?
Could help taking decisions for next step. Decision making is a barrier to execution. Suggestions of
decisions to the authorized employees for the decision making. Suggested decisions could be
approved or declined.

14. How will AI change the tools that R&D employees are utilizing?
Can be applied broadly in all tools not necessarily related to R&D, like traveling management tools,
purchasing tools. Could take care of more administrative tasks. Assist in understanding the R&D tools
(software) with minimal training.

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?
Yes, believe that it will be able to do this. Providing information from different sources, scientific

74
litterature and from different databases, related to protein structure. How chemical interactions can
be between them. Predict those interactions. AI could give suggestions of strongest (best)
interactions between molecules.

Questions related to “Technology” and “transform”:

16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?
Could help suggest new technologies for the people do decide on. Would bring more options than
what humans would do. Humans can become stuck in specific focus; AI can widen this.

Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer-centric innovation?
Involve the clinicians(customer), looking at it from a customer’s point of view. Looking at their needs.
AI would look into ways to create this innovation. Bottleneck could potentially be not being able to
ask/force customers to use AI.
18. How can AI contribute to the high growth markets and to the emerging markets?
Can help with the regulations in the different regions. To quicker interpret the regulations. This is
related to the regulatory department, but R&D also have to interpret this, depending on what
studies they need to do. Come to faster decisions.
19. How can AI contribute to delivering unique value proposition to customers?
Getting help quicker, in terms of service and support. Time is an important factor when releasing a
new product, could help what to focus on in the production and manufacturing. (forcasting
purporses)

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?
Could help to see the resources that we need at the moment. To understand what projects are in the
pipeline and interpret the resources that are needed. Can also help allocating resources between
groups. AI can potentially aid in deciding how to efficiently prioritize the work on the chosen projects.

21. How can AI applications in R&D influence the revenue?

To make the work more efficient, to help prioritize. To do more with the same resources.

22. How can AI influence the R&D investment?

Can interpret what is needed, could then both lead to increasing or lowering the budget. Could
predict potentials that needs to be developed and can therefore increase budget. It can influence,
but difficult to know exactly how.

23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)
Could potentially flatten the organization structure, with less levels.

75
Questions related to Transformational vs Transactional leadership skills:
24. How much do you involve your employees in decision making?
Quite a lot, concerning questions that are on the group level. More hands-on, not questions that are
for managers.
25. What do you think motivates your employees?
The ability to be able to affect their own situations. Impact the situation.

26. How do you react to mistakes/weaknesses of your employees?

If serious mistake, will take it in private. Talk about the mistake and how it affects the organization.
Would work with feedback. Making someone to learn from their mistakes.

27. Are you more oriented in the end result or the process in itself?
Currently more in the end result due to job situation but would like to be more focused on the
process itself.

28. How do you see AI implementation affects your leadership?

Some decisions will be taken without the involvement of emotions.

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI? (unasked)

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI?(unasked)

Interview date: 23/4/2021

Interviewee 6:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:

1. What kind of information can people in R&D get from AI?

2. Where does this information come from?

(Q1-Q2) You can get information from new research areas, about competitors launching tests in that
area, including publications about research, marketing information sent out by competitors. This is
useful for your development.

Also, you can go back deeply into the history for new development tests, AI can help quickly gather
information on what has been done before, both internal reports and external research reports done
by researchers.

3. How can this information be processed before being transferred to people?

76
Key words of interest should be imported by people and AI should investigate broadly and make
compilation of the information. AI needs to be trained iteratively to understand what the important
areas are, and finally come up with a development report based on which decisions are made.
4. Who/which roles will use this information?

Everyone will use this information, from people working with product development and customer-
related issues, and production deviations, post-marketing discussion on if test should be done or not.

Questions related to “People” and “transform”:

5. What new knowledge and skills are people required to have in using AI?

We should not need to have more skills if AI is good enough. People have the background knowledge
and expertise in the area but will need the help to find the information and compile it in a more
efficient way.

Further on, AI can also solve the problem for people e.g. choosing different ways to go with AI’s
imparalleled capacity in dealing with big data.
6. What changes are needed in the planning of employee development to use AI?
7. What changes are needed in talent acquisition?

(Q6-Q7)Not actually any changes needed. In my team, probably not needed. In large organization,
this can be affected because people would need to work differently, e.g. focus on other areas.

Generally, the acquisition of people would be the same, with the skillsets that we employ.
8. How do you think people's motivation is affected when working with AI?

If it is a good AI, the motivation would be positive.

Currently there is an information overload and this causes frustration and people find it difficult to
handle. Motivation will increase if there is a good tool in place to help with this.
9. How does AI change the knowledge sharing?

Not sure if knowledge sharing would be affected. It depends on how AI is set up, it can decrease the
sharing from person to person, if everything goes to AI.

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being applied?

11. How does AI support these processes?

(Q10-11) Generally all the processes should be included in AI applications, from the very beginning
initial information gathering, development, tests (AI can analyze the results by data input), until
product is finished.

12. What are the benefits?

77
Quicker in dealing with large amount of data, better overview and bigger picture of the gathered
information.

Quick probability analysis of different outcomes.

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed?

From the start of the project, we can get much more information than now, and we can have better
choices when we start the projects.
14. How will AI change the tools that R&D employees are utilizing?

It will decrease manual information handling.

Questions related to “Technology” and “inform and enable”:

15. Does AI provide information for understanding and predicting new technology trends? How?

Yes, it can gather so much data and it can see during time when has been the pattern and the trend
until now. If you have the history in a better overview, then it is easier for it to draw conclusion for
the future.

Questions related to “Technology” and “transform”:

16. How will AI change the way R&D decides on investment of developing or procuring new
technologies? (unasked)

Questions related to “Strategy” and “inform and enable”:

The strategy of Thermo Fisher Scientific:

17. How can AI contribute to customer-centric innovation?

AI can compile all the data about customer use of products and competitors’, and the trend analysis
and future possibility.
18. How can AI contribute to the high growth markets and to the emerging markets?

Maybe this is a marketing department question. AI will help us to be quicker to get new tests out in
the high growth, quicker if there is new customer need and help them quicker.

In emerging market, AI will spot out which trend will be good to go into and develop test for.
19. How can AI contribute to delivering unique value proposition to customers?

AI can hopefully make us quicker to react to customer, quicker delivery to customers. AI suggests the
future trends for the massive population need.

Questions related to “Strategy” and “transform”:

20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?

78
Hopefully AI will minimize the administrative burden and resources can have more time to work on
hands-on activities such as tests and development.
21. How can AI applications in R&D influence the revenue?

Much less manpower will be needed or put into development of new tests and delivery of new tests
to customers. Man hours are the highest costs.
22. How can AI influence the R&D investment?

AI will decrease the need in investment because less man hours will be used.
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)
It could affect the organizational hierarchy, by flattening the hierarchy because maybe some middle
layers would not be needed due to AI.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?

I involve them quite a lot, they are quite independent, and I strive to get them more independent so
that they do their own decisions.
25. What do you think motivates your employees?

Thermo Fisher is a great company that produces products that are important for the wellbeing of
people. We have good products that stimulate people to work with. The company strives to be a
really good employer.

26. How do you react to mistakes/weaknesses of your employees?

We have close discussion and follow-up meetings quite often and discuss around the problems and
see how we can learn from them.
27. Are you more oriented in the end result or the process in itself?

End results.
28. How do you see AI implementation affects your leadership?

It is a way to get information much more quickly and also to get it in a way that is easy to access. It
would be much more efficient in dealing with a large amount of data without missing important
details. It lowers the risk of missing unattended important information because AI has access to
wider networks to collect data, e.g. there are so many research publications every day which is
impossible for people to follow up and it takes a lot of time for individuals to keep track on it. This is
not only about research, but also legislation. This is something we could really have help from AI.

29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?

When a new project starts, we need as much background information as possible, and also to see
how the outcomes could be. The success recipe would be to able to get this information curated by
the AI from the messy amount of information, for decision making.

79
Also for predicting the outcome of multiple scenarios, AI can help out in simulating the results, not
only in product development but also in post-marketing e.g. how well the product is received by the
market and market share etc.

30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI?

Maybe the most challenging thing for now is to find the best AI tool to use and get it to work. We
have currently computer-based system but not with AI, which have difficulties to maintain to get
them to be used. If that is the problem also for AI, it would be very difficult to get them implemented.

Interview date: 29/4/2021

Interviewee 7&8:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:
1. What kind of information can people in R&D get from AI?
All kind of information.
AI is based on only information that you are feeding in, it depends on the training and the data that
you have fed in. Will not get sensible answers if asking outside the data that have been fed in.
2. Where does this information come from?
Depends what you want the AI to answer to you, depending on data gathering.
3. How can this information be processed before being transferred to people?
Important that it is enough of data and what kind of data it is. Processing is the second layer. Data is
more important.
Not only a calculation process, but also a business process. Meta-layer, starts with commitment of
higher management. Building machine learning system that gives narrow output. Collecting and
managing the data is a major undertaking, it is not very “easy”. Process starts from
the corporate level. Not a magical solution, much work behind. It is a lot of hard work.
4. Who/which roles will use this information?
Can help the development work, if it is done correctly. Enhance time of tedious hands-
on work. Could be part of the development work. Pipeline for the development work.
Most obvious application in an area with plenty of data available. In the development work that
is data-intensive. Applications related to much data. Data in the form of images, can use AI on.
Questions related to “People” and “transform”:
5. What new knowledge and skills are people required to have in using AI?
It is required in management. AI is not self-happening; much work needs to be done AI. People who
develop based on this data, or make decisions of this data need to have skills in understanding how
this is done.
Support from the different ranks, and management needs to be committed. Development team
needs to drive the process and manage data.
Also need the subject matter expertise that need to interpret the data/result and also judge if it
makes sense. AI can go wrong, therefore need someone to control.
6. What changes are needed in the planning of employee development to use AI?
Training is needed, and understanding and accepting. In people’s nature to resist changes.
Training, people have confidence in their self and their judgement. Need to understand the potential

80
but also the limitations of AI. Learning process for the humans. Learn where they can trust it. Need
both experience and training.
7. What changes are needed in talent acquisition?
Not many changes needed, always need to find talents. Potentially need to find people
experienced/knowledgeable with AI.
Difficult to find top notch experts. But in general, difficult to find talents.
8. How do you think people's motivation is affected when working with AI?
Different, both exited and not. Depends on how AI is programmed and its aim. Like any tool. Bad
tools are hated, and good tools become companions. Does technology really help us? Pushing people
to being more intelligent when taking routine tasks.
Positive effect, can take tedious routine tasks and give people more focus on other certain things
more related to human things.
9. How does AI change the knowledge sharing?
Challenge to share information between teams, not positive that AI can help.

Questions related to “Process” and “inform and enable”:

10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being
applied?
In all areas that have data. FDA and CE IVD, in regulated environments old fashioned methods do not
accept new methods easy. Slow process.
Can be implemented easier in internal processes then in external. Can be used in product. Will take
time. AI is a black box when used in products. Why? Need a lot of validation if used in regulated
environment.
11. How does AI support these processes? (unasked)

12. What are the benefits?

Data analyzed easier, help in time in processes with a lot of data. AI have to be devoloped in right
way.
AI is good (superb) with data that is difficult to combine in a traditional means. This is the strength of
AI. Better than our traditional methods.

Questions related to “Process” and “transform”:

13. How will AI change the way R&D work is being performed? (unasked)
14. How will AI change the tools that R&D employees are utilizing?
AI requires cloud based data management/storage.
Questions related to “Technology” and “inform and enable”:
15. Does AI provide information for understanding and predicting new technology trends? How?
No, we do not see how.
Questions related to “Technology” and “transform”:
16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?
It can change things, will affect other industries as well. Slower in the industry Thermo is working in.
Need a something that will initiate the change, some macro.
Questions related to “Strategy” and “inform and enable”:
The strategy of Thermo Fisher Scientific:
17. How can AI contribute to customer-centric innovation?
By helping developing new products. Faster launch of products. Can speed up.
18. How can AI contribute to the high growth markets and to the emerging markets?
Same as above, quicker. When much data, tedious to do the development. Speed things up.

81
19. How can AI contribute to delivering unique value proposition to customers?
Help to reduce manual work. Making decisions.
Questions related to “Strategy” and “transform”:
20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?
Decisions need to come from the corporate management. Will require recourses to manage the
data.
21. How can AI applications in R&D influence the revenue?
same as earlier, Being able to launch products faster and will have more products.
Can influence if a lot data from consumers is available. Difficult to use AI in data poor companies
(compared to google, FB and other).
22. How can AI influence the R&D investment?
Can save money if replaces employees.
In future, experts like doctors and lawyers, will be accompanied with different AI devices. Taking over
routine tasks. AI will reduce our need. Less need for certain jobs, like
less doctors needed because more effective with AI. More effective.
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)
When judgment is needed, humans are better. Other certain (admin related) task AI could help.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making? (unasked)
25. What do you think motivates your employees? (unasked)
26. How do you react to mistakes/weaknesses of your employees? (unasked)
27. Are you more oriented in the end result or the process in itself?
28. How do you see AI implementation affects your leadership?
Do not want and let it affect my leadership at all. Leadership requires social skills, AI not able to
replace this.
Could potentially affect leadership but current AI is far away from that. Routine work can be
replaced.
29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?
Need to have the correct ingredients (data, and understanding the mixing of ingredients).
Leadership team needs to be committed, and understand that it is not an easy process.
Different type of thinking.
Other: Risks with AI implementation. Risk of misinterpretation.
30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

Interview date: 30/4/2021

Interviewee 9:

Interview Questions related to AI applications vs R&D management activities:

Questions related to “People” and “inform and enable”:
1. What kind of information can people in R&D get from AI?
All kinds of data. We have huge datasets and we need to make our decisions and conclusions based
on them. The process of combining all the data from different datasets into
analysis for further decision making can be done by AI.
2. Where does this information come from?

82
Internal lab data for product improvement.
3. How can this information be processed before being transferred to people?
Just being processed through a software. Information of numbers and photos need to be analyzed
and reasoned. The data should also be verified for its correctness.
4. Who/which roles will use this information?
It would be a lot of departments involved when taking decisions, such as R&D, Marketing and
Production. R&D analyzes the data for product improvement and introduces to Production and
Marketing for decision making.
Questions related to “People” and “transform”:
5. What new knowledge and skills are people required to have in using AI?
If you are person that learns easily, no new requirements are needed. Same procedures that need to
be learned by human to conduct tasks will be used by AI. The thing is that you need to like
technology in the first place, since there are some skepticisms for some people who are in older
generations and like to do in the old way.
6. What changes are needed in the planning of employee development to use AI?
7. What changes are needed in talent acquisition?
(No.6-7) it is better to hire young people who are driven and eager to learn and can learn fast in
using the AI application.
For people who are already in the organization, AI introduction should be done softly, with regular
feedback from the people using it for improvements. Management also needs to hold a stronger
stand to push forward AI application by having a strong narrative to prove that it is for the future
success.
8. How do you think people's motivation is affected when working with AI?
If AI is introduced as a tool to improve people’s jobs and make the job easier and faster, and not
threaten people’s jobs, people will have good motivation.
9. How does AI change the knowledge sharing?
AI can bring better accuracy in knowledge sharing than human, and we can get more knowledge out
from the datasets. AI can bring consistency in analysis and avoid the emotion factors that human
has which can create erroneous analysis or conclusions. AI can detect some undetected aspects by
human.
Questions related to “Process” and “inform and enable”:
10. In which processes of R&D is AI applied? In which process of R&D do you think AI is being
applied?
The lab process could be improved by AI. Currently everything is done by hand, in this process AI can
be of help, in reasoning of problem and problem solving.
11. How does AI support these processes?
Withdrawing data and data analysis parts are those where AI can be applied. Suggestions will be
given by AI but human should be the decision makers.
12. What are the benefits?
Making the process faster and more accurate in data analysis.
Questions related to “Process” and “transform”:
13. How will AI change the way R&D work is being performed?
The whole process will be faster from months to hours and more accurate, and interpret some
information that could be missed by human.
14. How will AI change the tools that R&D employees are utilizing? (unasked)
Questions related to “Technology” and “inform and enable”:
15. Does AI provide information for understanding and predicting new technology trends? How?
Yes, the “black box” AI can take out conclusions from datasets which human cannot understand the
reasoning part. E.g. for drug development, AI has better predictability in choosing the right drug than
clinical studies.

83
Questions related to “Technology” and “transform”:
16. How will AI change the way R&D decides on investment of developing or procuring new
technologies?
Some groups have higher workload and some have lower. People need to sit down to understand
how much work would be eased by AI. Definitely we should invest more in AI in R&D.
Questions related to “Strategy” and “inform and enable”:
The strategy of Thermo Fisher Scientific:
17. How can AI contribute to customer-centric innovation?
For product improvement, AI can be faster and more accurate, this can make customer more
satisfied with better products.
18. How can AI contribute to the high growth markets and to the emerging markets?
The product I am working with is working at the moment but soon to be outdated. With the help of
AI, we can have better product in a much faster pace. We can test new ideas faster.
19. How can AI contribute to delivering unique value proposition to customers? (unasked)
Questions related to “Strategy” and “transform”:
20. How will AI affect the decision making for R&D resource (money, people, technology) allocation
and prioritization?
With AI application, maybe R&D department could be visualized in terms of different workloads in
different groups, which helps re-allocate human resources.
21. How can AI applications in R&D influence the revenue?
Only in good ways. AI can work 24/7, faster and more accurate and avoid human errors and
situations.
22. How can AI influence the R&D investment?
The investment will go up. We need to buy the technology and integrate it to R&D. The time and
labor investment will go up in the development of the AI applications.
23. Does AI influence organization structure? How? (More integration? centralized or de-centralized)
It would influence especially production people if AI would be used. R&D and QA, marketing as well
in different ways and different amounts. But Production and R&D will be more influenced because it
is more process-driven and AI can replace this part.
In R&D there will be less humans. Marketing will not have as much reduction as R&D.

Questions related to Transformational vs Transactional leadership skills:

24. How much do you involve your employees in decision making?
Employees are rather involved but not enough, especially on bigger decisions. We could be much
more involved in big decision making. It is often that the employees who understand the
improvement areas and what should be fixed. Management should listen more
25. What do you think motivates your employees?
It depends on what type of daily work the employees are working with. You need to be satisfied with
what you are doing. Salary and leadership such as what kind of support you can get are very
important. Colleagues play a role in motivation as well; you need to feel comfortable and fun with
your colleagues at work.
26. How do you react to mistakes/weaknesses of your employees? (unasked)
27. Are you more oriented in the end result or the process in itself? (unasked)
28. How do you see AI implementation affects your leadership?
There are a lot of improvements which AI could help with. My work is very hand-based, and
everything is done by human. AI has not been implemented in our system. We are doing the whole
process manually, from handling chemical components, to evaluate SA and draw the conclusions
from the data. Sometimes the conclusions are wrong so that process needs to be repeated. This is a
very time-consuming process and AI is of great use in doing these steps instead, in doing the SA by

84
interpreting the data and reasoning, and getting best conclusions for us. AI can make the whole
process faster and more accurate.
29. What do you think is the ideal recipe to achieve success in R&D innovation with AI?
It should be clear that AI should just be a digital assistant and would not replace human and we are
not going to lose our jobs. It should be introduced as a tool to make our jobs easier.
30. What do you think are currently the most critical improvement areas in R&D management in the
age of AI? (unasked)

85
Appendix 3: Diagram of 1st Order Concepts and 2nd Order Themes
1st Order Concepts 2nd Order Themes
"Inform&Enable" and "People"
-AI provides information about product users, markets and operations via figures and text from customers through behavior pattern recognition, helping development with safety, ergonomics and simplicity.
- Information sources can come from users using the products, from new research areas about competitors' products including publications about research, market information. Quicker and broader access to
- Historical information and knowledge can be retrieved deeply by AI retrospectively quickly from internal and external report systems by researchers for product development.
internal and external information
- Information should be processed by AI similar to data management system to read and categorize information
- AI should investigate broadly and compile big data with people's specific aim, and should be trained iteratively to understand what the important areas are, and finally give suggestions based on which decisions are made. (customers, market, competitors,
- AI, with clear defined aims, captures the "pattern" which is the result of its data analysis and shares with people. operations, historical data and
- Information receivers can be product designers with hands-on skills, creativity and entrepreneurial mind; both employees at basic levels and as R&D managers, from all positions from product development, customer-related issues, production deviations, as well as post-marketing. scientific research)
-Information can come from instruments, software, technologies, processes previous studies and experiments.
-AI output should be validated and evaluated by the employees.
-Processing is needed, to be able to be relevant to employees.
-Both employees (lab people) and managment team will use this information. Simpler and more objective in
-Plenty of information is available for AI, software can be helpful.
-Information can come from anywhere, this depends on what we are looking for. Could come from customers, surveys from end users and from the critical mass. analysis and more efficient decision
-Difficult to process information, streamlining needs to be concidered and what is of interest for us. Need to concider what is possible and feasable. making
-Funneling information is needed before it reaches people.
-Managers will definitly use this information (not limited to R&D)
-Information should be processed though a comittee composed of different departments. Information should flow in the right order.
-Manufacturing also important and relevant.
-Data for data analysis, but also for customer experience. Different way for getting data.
-Spot-on products, information of how they might want to use the products.
-AI can help with big data for product devolopment. Purposeful data processing needed
-Data comes from customers.
-Data needs to be visualised in a visual and easy to understand way. Processed data.
-Time is limited for managers and they are exposed to much information.
-AI can be used in different roles (multiple levels)
-AI information can be used to validate the product in the end of devolopment phase. Can also be used to show that the product is safe for authorities to approve.
-AI information will be used in the beginning of product devolopment phase by employees (to understand which kind of product to devolop.
-AI will be used it the end of the producs devolopment cycle to understand what to get out of the product. General applicability to managers
-AI information can aid during the process, can revisit the different phases.
and employees at lower levels
"Transform" and "People"
- AI requires people to have new skills such as analytical skills on data quality and reliability.
- No new skills will hopefully be needed since AI's exceptional capacity of processing large data can assist us, while possibily simulating outcomes from multiple scenarios for people to select the best option.
- Nothing needs to be changed should remain the same, with the same skillsets required.
- Leaders should understand the needed skills and how to make the changes, in order to plan for employee development and talent acquisition
- Training and education are crucial for people to be onboard in using AI. Questionable change in the
- The way to "sell" AI to employees is important to avoid demotivation: clear purpose, benefits, and how to succeed. required skillsets for people
- General media's coverage on AI, generational preferences, and how the communication is done are important for people's motivation to use AI.
- If the AI used is good, motivation will increase since there is an information overload and this causes frustration in handling the massive data manually.
- AI gives stronger argumentations with more objectivity for people
- Depending on how AI is set up, knowledge sharing could be decreased from people to people, mainly due to more interactions with AI.
-People need an open mind for AI implementation.
- People also need more IT and computer skills, to be able to understand the information and interpret the result. Neccesary trainings in usability,
- Training on AI methodology for gathering information, to be able to be more observant when controling and validating the result. Instead of training of the technology related to the specific role, more training will be needed for the AI.
-Component people will still be important (experts in the field), however AI skills would be necessary. If AI will replace tasks related to specific skills, those skills will not be needed anymore. interpretation and change
-The effect on the motivation will depend on what AI is being used for. E.g. if used for quality improvement, will have a positive effect and if used to replace peoples tasks and jobs, this will have a negative effect. management
-The effect on the motivation will depend on the employees interest.
-Technical skills of how to use AI needed and skills in data analysis.
-Technically challenging to get people with high skills and knowledge, but training to understand AI data would not be difficult.
-AI’s purpose is to make our lives easier.
-People with different skils for diversity.
-Not to forget people with experience.
Updated employee development
-Creativity and experience meet to make things work. E.g. Person from travel agency started in Thermo, no experience in life science but experienec in digital marketing. plan and talent acquisition strategy
-Positive impact on motivation, new, exiting and trendy field with much progress. for managers
-AI can make it quicker to find solutions in regards of data sharing in the organization.
-new improvement ideas, and time saving when sharing between groups.
-24/7 sharing can aid knowledge sharing in different geographies.
-Need skills to be able to make use of AI technology (engineering skills).
-Need skills to be able to interpret the AI data.
-Enginners that have devoloped the AI needs
-Long plan is needed for the AI implementation and employee devolopment. R&D is conservative.
Good motivator if communication is
-Both managers and followers needs to be driving this change to AI. well and AI is useful
-Motivation will be mixed, whole spectrum. Will depend on generation.
-AI can stimulate collaboration. Bring different departments together, seeing the bigger picture.

"Inform&Enable" and "Process" More collaboration with different

- Generally all the processes in R&D should be included in AI applications, from the very beginning of initial info gathering, to development phases, tests, until product development is finished. departments
- Regarding benefits to R&D processes, AI can provide quicker data handling from big data, giving better overview and a more complete picture of different information, with also possibility in evaluating the outcome probability of different scenarios.
-AI can be applied in both in laboratoty work and in managment. E.g. Project managment, processes, and technology in the lab.
-AI can support onboarding and simplify this, specific learning for different roles/people. Automate certain steps.
-The benefits might be knowledge sharing between different processes in managment, more personalized decisions. Saving time moving to the next step (risk analysis and change request). Processes can be simplified by taking decisions on the next step.
-data collection part to justify the project.
-All processes are important (early development process, validation process, manufacturing, quality assurance)
-AI is easier to implement in devolopment then in pure research. AI to be implemented in different
-AI collect imput requirements, help to process more, more data. Reach more broader then without. processes of R&D.
-AI can help track the progress, close down projects that shouldn continue.
-AI Benefit is the devolopment products without feelings involved. "someones baby",
-AI will influence the decision making. Related to bias when making decisions.

"Transform" and "Process"

- AI provides objective data-based analytics for R&D management in decision making in product design with different information from e.g. customers, operators and engineers AI valuable/beneficial in processes
- From the start of the project, more information can be obtained and better choices can be selected when projects are initiated.
related to data collection and big
- With the aim to develop products faster with lower costs by using AI in the long run, there can be a shift that products produced in high labor cost markets would be produced in countries with low labor costs, which leads to a geographic shift of R&D activities.
- AI can change R&D in decision making for the next step. Decision making can be a barrier to execution. AI could suggest examples of decisions for the employees to chose from. data.
- AI can aid general tools also, not only related to R&D tools. Travel managment, purchasing tools. Could take care of administrative tasks. For R&D tools, it could aid in understanding the tools with minimal training.
-QUciker and more efficient with AI.
- Easier to make decisions and move forward.
-Helpful in detecting errors and monitoring processes.
-Quality control.
-AI can Influence the decision process. Better analysis and reduce studies. Can Extrapolate better and avoid repeating studies.
-AI can do more simulations, and therefore less tests.
Streamlining and simplifying
processes.
"Inform&Enable" and "Technology"
- It would be a challenge that AI is smart enough to understand and predict new technology trends.
- Based on big data processing, AI can see pattern and trend. Better overview based on historical data, it is easier for AI to draw conclusion for the future.
-Provide information from different sources, scientific literature and from different databases. Could for example give suggestions of chemical structures and bonds between them.
- -We can streamline what we are interested in by using AI by filtering the noise.
-AI can simply things by e.g. adding new possibilities or eliminating things that do not give added values. Better predictability and
-It can be used to predict new technology. Maybe more then we think. It can pick up trends quicker.
suggestions to human
"Transform" and "Technology"
- AI can possibly bring changes in the way R&D invests new technology to produce new products, while AI replaces physical lab and reduces operational and labor costs.
-Could suggest new technologies for people to decide on and would bring more options with a wider view.
- R&D will see the opportunities and benefits with AI. R&D will move forward because AI gives opportunity for easier access and easier ways of working in the future. There will be interest in investing in this area. More customer-oriented and better
-AI can predict or be one way of predicting, same methods will probably still be used but AI will not involve emotions in the process. value proposition by more efficient
customer interactions and more
precise understanding of customer
"Inform&Enable" and "Strategy"
- AI contributes to customer centric innovation by informing R&D management how users use products, how to provide them with faster and safer solutions, how to train customers in using the product and give suitable suggestions to customers to help their workflows.
needs
- AI can contribute to user data of competitors' production, and do the trend analysis for seeking future opportunities.
- AI can help us be quicker to get new products in the high-growth market and help customers in need quicker. AI can spot out the right trend in emerging markets to develop products for.
- AI contributes to delivering unique value proposition to customers by making more personalized products with more information on customer needs and their usage of products, and more understanding on customers' pain points.
Quicker growth in high-growth and
- AI contributes to suggesting the future trends of the need of massive population, not just individuals.
- AI contributes to quicker adapatation of customer needs and quicker delivery to customers. emerging markets with better
-Involve the customers and looking at it from customers point of view, looking at their needs. intelligence in demand, competitors
-AI would look into wais to create this innovations based on customer needs. and customer behavioral patterns.
-AI could speed up the process and get faster decisions, based on for eg different regulations in different markets (EM and HG markets). Can quicker interpret the regulations.
-Use AI to inform customers of our offerings, and how it would benefit them.
-High grow markets, new things and technologies are taken in a faster pace.Early adopters. Keen to jump into using new technology to simplify and make things easier.
-Slower adoptation in emerging markets.
- AI can provide new value proposition, unique selling points by showcasing them to the customers
-AI can help evaluate and simulate customers and get the bigger user base. It is costly to go out to customers.
Flatter organizational hierarchy
- AI can help with the analysis of market needs, marketing analysis about the customer needs.
-AI could also accelerate development and market access of a product
-EM are difficult to get (our) products in to. If could get a better understanding of what we need there, it could accelerate growth.
-AI could help to develop the value proposition, to make that more powerful.

"Transform" and "Strategy"

- Depending on the strategic role of AI the company is positioning, AI provides decision makers with stronger ground and more informative and reliable suggestions in their decision making.
- AI can minimize administrative burden and people have more time to work on practical activities.
More efficiency
- AI can analyze big data to come up with suggestions on management decisions on prices, locations and timing to produce. (quicker/faster/cost)
- AI can influence revenue positively by providing R&D with faster decisions and adapting customers needs quicker, and possibility to shift faster before the real shift occurs, as well as by reducing costs and increasing quality.
- AI can increase revenue by the reduction of man hour cost put in the development of new product and delivery to customers.
- AI can provide stronger evidence on which investment and how investment should be made.
- AI can decrease the monetary need in investment due to the decrease of manpower cost which is the highest cost in R&D
- AI can replace some manual work and ways of doing things. But people should have same roles with extended assistance of AI with faster and more accurate information.
- Organizational hierarchy can be affected by AI to be flatter because there is possibility that some middle layers' tasks could be replaced by AI.
-Predict the resources needed, based on the projects that are in pipeline.
-Can help allocating resources between groups. Guaranteed management support
-Efficient prioritization of the work for the chosen projects.
-Can both increase and decrease the budget, depending on AIs interpretation.
- Could potentially flatten the organization structure, with less levels.
-Data that is being captures and presented to higher managment.
-AI can showcase the summarized information to GLT or decision makers, which makes it easier for managers to explain the logics behind. We do not need to show all the data points but only the important ones which AI helps you in identifying.
- AI can interpret information for management from massive data.
- AI can help revenue increase in different positions: sales, new product development, etc. AI can close the information loop by feeding back information from one end to another end. In this way it can help bring more revenue to the company.
Sufficient and correct technology
- When you need to make a new investment it will need massive data and justify all the data which AI can provide its help with, in an understandable way. The database is big and the significance of the data will be high and validity is therefore high.
-A lot of processes will be changed, and will be easier. investments
- Maybe certain types of job will not be relevant for the future, meaning certain roles will be replaced by AI.
- It will not happen overnight but gradually. Companies should educate people in the areas which can be replaced by AI to work in other areas so that they can work in somewhere different.
-AI in decision making will be a factor in the future, AI will support the decision but providing information and analysis.
-AI will not be able to replace us, human touch. But will be a factor we take into concideration.
-AI will get more out of the organization. By using good tools, with AI and other tools, and drive efficiency.
-R&D investment will take longer time to implement AI in.
-AI will influence certain things in the org structure, e.g. service centers (complaints and rutine things). Big challenges in finding the most
-AI could replace people at certain levels, but more "difficult" things need to be handled by people.
-Data is already available but not really accessible for R&D (from marketing e.g.). AI can help here.
suitable tool and implementing it

AI Impacts on R&D Leadership

- AI can help assist leaders in product development with speed and safety
- There is an imminent need for leaders' agility to start getting understanding on how AI can help them, which skills are needed.
- AI provides a way for leaders to get information much more quickly and get a better overview from the enormous data access with AI's exceptional capability to process and analyze data. Leaders will have lower risks of missing important details or information from the massive data which AI has
access to. Research publications and new regulations are being updated every day and it is impossible for people to follow-up, and AI can help with these with high objectivity, accurary and speed.
-Decisions will be taken without the involvement of emotions.
-Saving time and bring more efficiency
-Have not seen it so far in my leadership but believes it will come, in the future. Will be an additional tool for decision making in the future.
How to Achieve Successful R&D innovation with AI
- Specific support, sufficient investment in technology and people who are open for changes, where change management is crucially pivital throughout the organization.
- Technically, AI needs to have the capability to give as much background information as possible and simulate the outcomes of different scenarios, and then curate the information from huge amount of data to provide suggestions to help people make decisions.
- AI should be used in many ways, not only for product development, but also in other functions such as post-marketing to understand how well the product is received by market and also market share.
- The most challenging thing for getting success in using AI can be to find the best AI tool and get it working. The big difficulties lie in maintaining it and get it working as it is desired to be in implementation.
-AI can streamline the R&D and innovation.
R&D instrument, evaluate the footprint before the actual product. AI can aid.
- AI can help indicate if to move forward with a R&D project.
-The analysis part, tools that can draw conclusions quicker then we can today. Where I am working in, it is mainly data analysis. Believe that this will differ in other areas.

Details are presented in attachment “Appendix 3_Diagram of 1st Order Concepts and 2nd Order
Themes.pdf”.

86
Appendix 4: Diagram of 2nd Order Concepts and Aggregated Dimensions

2nd Order Themes Aggregated Dimensions 2nd Order Themes

1.Quicker and broader access to

internal and external information
(customers, market, competitors,
11. Necessary human control in AI
operations, historical data and
outputs
scientific research)

2. Simpler and more objective in

1. Shift in leadership 12. AI valuable/beneficial in
analysis and more efficient decision elements processes related to data collection
making and big data.

13. Streamlining and simplifying

3. Purposeful data processing
processes.
needed for ethics, relevance, 2. Shift in employees'
interpretation, correctness and (managers and followers)
information flow direction
needed skills
14. Better predictability and
provision of multi-scenario
suggestions to human

4. AI information generally 3. Organizational 15. More customer-oriented and

applicable to managers and integration (streamlining & better value proposition by more
employees at lower levels flattening&less silo) efficient customer interactions and
more precise understanding of
customer needs

16. Quicker growth in high-growth

4. Management strategy
and emerging markets with better
5. Questionable change in the forming aid intelligence in demand, competitors
required skillsets for people
and customer behavioral patterns.

6. Neccesary trainings in usability 17. Replacing certain human

(IT skills), interpretation (statistics) 5. Objectivity by avoiding elements (e.g hard skills) and flatter
and change management emotions and biases organizational hierarchy

18. Higher efficiency (quicker/more

7. Updated employee development
accurate/cost)
plan and talent acquisition strategy
for managers
6. Efficiency booster
19. Guaranteed management
commitment and understanding of
8. Good motivator for employee if AI limitations for innovation
communication is well and AI is success
practically useful

7. Innovation stimulator
20. Sufficient and correct
9. More knowledge sharing and technology investments
collaboration with different
departments in most cases

8. Profit and growth driver 21. Big challenges and needs in

finding the most suitable tool and
10. AI to be implemented in
implementing it
different processes of R&D.

Details are presented in attachment “Appendix 4_Diagram of 2nd Order Concepts and Aggregated
Dimensions.pdf”.

87