AI and Firm Performance

Journal of Business Research 174 (2024) 114500
Contents lists available at ScienceDirect
Journal of Business Research

journal homepage: www.elsevier.com/locate/jbusres
Artificial intelligence and adaptive response to market changes: A strategy

to enhance firm performance and innovation
Yulia Sullivan a, Samuel Fosso Wamba b, *
a
Hankamer School of Business, Baylor University, One Bear Place #98005, Foster Campus, Waco, TX, USA
b
TBS Business School, France 1 Place Alphonse Jourdain, 31068 Toulouse, France
A R T I C L E I N F O A B S T R A C T
Keywords: This research examines how AI-powered capabilities can bring value to organizations by enhancing their
Artificial intelligence Adaptive Response to Market Changes (ARMC). Utilizing insights from organizational agility and the dynamic
Adaptive response to market changes capability framework, we define ARMC as an organization’s ability to promptly identify and adjust to market
Innovation
changes, with customer responsiveness and operational adjustment as foundational competencies. We outline
Firm performance
AI-powered capabilities
three AI-powered capabilities (AI-enabled automation, AI-enabled analytics, and AI-enabled relational capabil
ities) as ARMC’s predictors. We posit that the strengths of these relationships depend on environmental hostility
and dynamism. Additionally, we propose positive associations between ARMC and three organizational out
comes: firm performance, process innovation, and product innovation. Our research employs a two-stage design,
surveying IT and business executives from firms that have adopted AI. The results demonstrate significant
interaction effects of environmental hostility and dynamism on the relationships between AI-powered capabil
ities and ARMC. Furthermore, we find that ARMC positively influences firm performance and innovation.
1. Introduction introduces a new set of challenges to this longstanding issue. Conse

quently, more contemporary scrutiny and analysis is needed (Nambisan
The rise of digitization, characterized by the profound use of data et al., 2017). The emergence of new AI capabilities and their associated
analytics and artificial intelligence (AI), has compelled organizations to challenges raises the question of how firms can effectively harness these
adopt a new perspective on their IT assets and focus on a more intricate technological advancements to gain a competitive edge in their in
level (Benner & Tushman, 2015). Typically defined as “the ability of dustries (Iansati & Lakhari, 2020).
machines to perform human-like cognitive tasks,” AI is currently viewed Although prior studies have shown the promising role of AI in sup
as the most important and disruptive new technology for organizations porting decision-making processes in an organization (e.g., Duan et al.,
(Benbya et al., 2020, p. ix). However, the management of AI differs 2019; Gupta et al., 2022; Jussupow et al., 2021; Shrestha et al., 2019),
significantly from previous approaches to IT management (Berente theoretical consensus is limited on how AI influences firms’ responses
et al., 2021). AI cannot be viewed merely as a technology or a collection and adaptability to market changes. However, AI can empower orga
of technologies; rather, it represents an ever-evolving system encom nizations to respond to dynamic market changes effectively (Narayandas
passing emerging computational capabilities (Berente et al., 2021). & Sengupta, 2023). For example, it can automate and optimize various
Thus, today’s leaders must navigate both the opportunities and the ob operation processes, enhancing the efficiency and responsiveness of an
stacles that come with the pervasive adoption of AI. organization (Benbya et al., 2020; Huang & Rust, 2021). AI-enabled
Further, the process of digitalization enabled by AI has triggered a predictive analytics allow organizations to identify emerging market
paradigm shift in the fundamental theories that explain how IT inno trends and connect with customers, allowing them to proactively adapt
vation generates value (Nambisan et al., 2017). While the challenges of their strategies and offerings (Huang & Rust, 2021). Moreover, by
utilizing IT to create business value are fairly well understood (Barua leveraging AI, businesses can gain valuable insights from vast volumes
et al., 2004; Côrte-Real et al., 2017; Grover et al., 2018; Kohli & Grover, of data and make informed decisions in real time (Davenport et al.,
2008; Melville et al., 2004; Wang et al., 2012), the recent rise of AI 2020). In sum, AI provides organizations with advanced computing
* Corresponding author.
E-mail addresses: yulia_sullivan@baylor.edu (Y. Sullivan), s.fosso-wamba@tbs-education.fr (S. Fosso Wamba).
https://doi.org/10.1016/j.jbusres.2024.114500
Received 10 March 2022; Received in revised form 29 December 2023; Accepted 4 January 2024
Available online 13 January 2024
0148-2963/© 2024 Elsevier Inc. All rights reserved.
Y. Sullivan and S. Fosso Wamba Journal of Business Research 174 (2024) 114500
abilities to perform mechanical tasks (i.e., AI-enabled automation capa In the context of ARMC, organizational agility emerges as the most
bility), process information and complete analytics tasks (i.e., AI-enabled closely related concept, encompassing companies’ ability to flexibly
analytics capability), and directly interact with humans (i.e., AI-enabled adapt to shifting market conditions and seize opportunities for sustain
relational capability) (Huang & Rust, 2021a, 2021b). Thus, these able growth and competitiveness. In their recent work, Kalaignanam
computing abilities help companies generate new ways to enhance their et al. (2021) explored the commonalities and distinctions among dy
responses to market opportunities. namic capabilities, agility, and associated concepts. According to their
In this study, we explore how AI-powered capabilities can create findings, dynamic capabilities encompass sensemaking, which involves
value for organizations, leading to enhanced firm performance and firms’ ability to respond to unexpected events by discerning the events
innovation. These capabilities are crucial in establishing firms’ ability to unfolding and striving to establish a more structured environment, and
swiftly identify and respond to market changes, such as shifts in speed, which represents the swiftness with which firms sense changes
customer preferences or new market trends. We refer to this concept as and initiate responses. However, dynamic capabilities lack the element
“Adaptive Response to Market Changes” (ARMC). Our emphasis is on of iteration. In contrast, agility includes the critical aspect of iteration,
adopting a core competencies perspective (Ravichandran et al., 2005) to entailing the iterative refinement of decisions before their relaunch or
effectively reflect how firms respond to market changes in two key areas: scaling to minimize uncertainty. Thus, while agility and dynamic ca
customer responsiveness and operational adjustment. pabilities overlap, they also have distinct features, such as agility
Furthermore, environmental conditions influence firms’ capability- emphasizing the iterative nature of decision-making to reduce
building processes and responses to market demands (Lee et al., uncertainty.
2015). However, limited research has explored how these conditions The dynamic capabilities framework also recognizes the importance
affect firms’ responses to market changes concerning AI even though of agility. However, it posits that the continual pursuit of agility may not
they play a crucial role in shaping managerial decision-making (Pavlou always be necessary since doing so could lead to a loss of efficiency and
& El Sawy, 2011; Wilden & Gudergan, 2015). While strategic decisions additional costs. Instead, agility becomes imperative primarily when
in product or market adaptation can be unpredictable in rapidly market changes are linked to uncertainty or risks (Teece et al., 2016).
changing environments (Eisenhardt & Martin, 2000; Mitchell et al., Put differently, firms may not require constant agility except in situa
2011), our study proposes that the effects of AI-powered capabilities on tions where uncertainty is present. Consequently, when firms encounter
ARMC depend on specific environmental conditions. Specifically, we uncertainty, they can enhance their agility by leveraging dynamic ca
theorize that two environmental conditions—environmental hostility and pabilities. Next, we explain ARMC from the agility and dynamic capa
environmental dynamism—affect the degree to which AI-powered capa bilities framework.
bilities drive firms’ responses to market changes.
Focusing on this objective, we will elucidate how the interplay be
tween AI-powered capabilities and environmental conditions can result 2.2. Adaptive response to market changes (ARMC)
in different levels of ARMC, which directly affects firm performance,
process innovation, and product innovation. Our study seeks a deeper We derive the concept of ARMC from the organizational agility
understanding of the circumstances under which AI-powered capabil perspective. Scholars have offered various definitions of agility
ities will likely yield benefits or drawbacks for organizations. Moreover, depending on the context. For example, in the marketing field,
our investigation into how AI-powered capabilities facilitate ARMC Kalaignanam et al. (2021) conceptualized marketing agility as “the
contributes to AI research beyond its purely technical aspects (e.g., extent to which an entity rapidly iterates between making sense of the
Asatiani et al., 2020). Thus, the empirical insights from this study offer market and executing marketing decisions to adapt to the market” (p.
practical guidance to organizations, enabling them to anticipate and 36). Meanwhile, in the information systems (IS) field, Tallon and Pin
prepare for the economic impacts of AI implementation. sonneault (2011) conceptualized organizational agility as “the ability to
detect and respond to opportunities and threats with ease, speed, and
2. Theoretical background dexterity” (p. 464). Sambamurthy et al. (2003) also proposed that agility
comprises three interrelated capabilities: customer, partnering, and
2.1. An overview of dynamic capabilities operational agility. Moreover, Lu and Ramamurthy (2011) conceptual
ized two dimensions of organization agility: market capitalizing and
Dynamic capabilities are defined as “a firm’s ability to integrate, operational adjustment agility.
build, and reconfigure internal and external competencies to address Organizational agility, as highlighted by Lu and Ramamurthy
rapidly changing environments” (Teece et al., 1997, p. 516). Dynamic (2011), represents a comprehensive firm-wide capability to effectively
capabilities are distinct from operational capabilities because they serve navigate through unexpected changes in dynamic business environ
different organizational functions. While operational capabilities are ments. Given the breadth of this concept, encompassing all its elements
essential for companies’ smooth functioning in day-to-day operations, within a single study is challenging. Nonetheless, in our research, we
dynamic capabilities help organizations achieve and sustain a compet focus on two crucial elements related to organizational agility that
itive advantage (Lin & Kunnathur, 2019; Teece, 2007; Teece et al., ARMC1 captures: customer responsiveness and operational adjustment/
1997). Firms demonstrate dynamism in capabilities by adapting strate adaptiveness dimensions.2 These two dimensions specifically focus on the
gies to changing market conditions and creatively combining existing two core competencies found within organizations, namely market-
resources in innovative and unconventional ways (Morgan et al., 2009).
Dynamic capabilities constitute firms’ ability to (1) sense and shape
1
opportunities and threats, (2) respond to market opportunities, and (3) Instead of utilizing the term “organizational agility,” which encompasses a
maintain competitiveness through enhancing, aligning, and reconfi firm-wide capability, we opt for the more precise term Adaptive Response to
guring firms’ intangible and tangible resources (Teece et al., 1997). Market Changes (ARMC) to refer to the specific focus of our study. We thank the
Handling Editor for this suggestion.
Although the primary idea of dynamic capabilities centers on these 2
Utilizing Kalaignanam et al.’s (2021) criteria for agility, we contend that
sensing, seizing, and reconfiguring activities (Teece, 2007), scholars
customer responsiveness entails marketing decisions, especially those related to
have proposed several other dynamic capabilities, including dynamic customer or product choices, in addition to sensemaking and speed. On the
managerial capabilities (Helfat & Peteraf, 2015), organizational agility other hand, operational adjustment involves sensemaking and speed. Although
(Chakravarty et al., 2013; Nazir & Pinsonneault, 2012; Sambamurthy we acknowledge that the iteration component of agility is difficult to assess in a
et al., 2003), market orientation (Morgan et al., 2009), and new product cross-sectional study, we assume that for decision-makers to arrive at market
development (Pavlou & El Sawy, 2006). decisions, they must adapt and refine the problem.
2
access and functional-related competencies (Battleson et al., 2016; Table 1

Bradley et al., 2012; Chen et al., 2017). Conceptualization of AI-powered capabilities.
Market-access competencies encompass all activities that enable AI Capability Definition Examples
firms to connect with their customers and effectively identify their needs
AI-enabled AI’s capability to do Using AI to get information;
(Ravichandran et al., 2005). In the context of ARMC, customer respon automation mechanical and repetitive using AI to perform general
siveness represents a crucial aspect of market-access competencies, capability tasks with relatively limited physical activities; using AI to
encompassing companies’ ability to actively engage with customers to amounts of learning or perform administrative
foster collaboration, explore innovative opportunities, and develop adaptation (Huang et al., activities (Huang et al., 2019).
2019)
competitive strategies in response to market changes (Sambamurthy AI-enabled AI’s capability to analyze and Using AI to personalize
et al., 2003). For instance, customer responsiveness could include analytics make decisions based on vast products based on customers’
adapting to shifts in aggregate customer demand and customizing capability volumes of data (Davenport & preferences; using AI to
products or services to cater to individual customer preferences. Ronanki, 2018) anticipate credit card fraud;
using AI to predict equipment
Operational-related competencies encompass all internal business pro
failures (Davenport &
cess activities that enable firms to deliver distinctive and dependable Ronanki, 2018).
products while enhancing the speed and precision of key operations AI-enabled AI’s capability to recognize, Using AI bots to provide sales
(Chen et al., 2017). Within these competencies, operational adjustment relational emulate, and respond and customer service supports;
assumes significant importance. Operational adjustment refers to firms’ capability appropriately to human using virtual agents to assist
requests by learning and and care for others (Huang
ability to promptly modify their internal business processes, facilitating adapting from experience ( et al., 2019).
effective adaptation to shifts in market demands (Lu & Ramamurthy, Huang et al., 2019)
2011). For instance, operational adjustments could involve expanding
into a new market and adjusting the range of products or services offered
for sale. described as “analytics on steroids” (Davenport & Ronanki, 2018, p. 4).
AI with this capability generally uses machine-learning algorithms to
2.3. Artificial intelligence and AI-Powered capabilities detect patterns and interpret their meanings in decision-making pro
cesses (e.g., predict what a particular segment of customers is likely to
AI is “the frontier of computational advancements that references buy and analyze warranty data to identify the safety problems in auto
human intelligence in addressing ever more complex decision-making mobiles and other manufactured products) (Huang et al., 2019). Unlike
problems” (Berente et al., 2021, p. 1435). Currently viewed as the traditional analytics, AI-enabled analytics involves learning and adapt
most important and disruptive new technology for organizations (Ben ing systematically from data (Huang et al., 2019). An AI system can
bya et al., 2021), AI is revolutionizing managerial practices for process both unstructured and structured data, as well as information
analyzing performance and developing novel insights (Kaplan & Haen generated by the system itself to acquire new or modify existing
lein, 2019). In addition, it includes statistical machine learning, neural knowledge (Paschen et al., 2020). Lastly, AI-enabled relational capability
networks, and natural language processing (NLP) (Davenport & engages with human users using NLP chatbots, intelligent agents, and
Mahidhar, 2018). machine learning (Davenport & Ronanki, 2018). AI with a relational
Technology is widely acknowledged as a fundamental resource that capability allows users to directly engage with a virtual agent rather
organizations can effectively exploit, deploy, and leverage to their than with human customer-service agents (Davenport & Ronanki,
advantage (Manis & Madhavaram, 2023). In light of the capability- 2018). Such AI can learn and adapt its behaviors by learning from its
building process, the significance of investments in IT lies in its suc interactions with human users (Huang & Rust, 2021a).
cessful integration into the core activities of organizations (Doherty &
Terry, 2009; Ravichandran et al., 2005; Sambamurthy et al., 2003). In 3. Research model and hypotheses development
this context, Information Systems (IS) serves as a crucial mechanism in
establishing coherence between IT activities and firm priorities, thereby 3.1. Research model
facilitating the development of comprehensive capabilities that enable
seamless cross-functional integration and coordination (Chen et al., Fig. 1 presents our conceptual model. In proposing our research
2017; Ravichandran et al., 2005). Through this lens, our focus centers on model, we posit that the three AI-powered capabilities act as predictors
three AI-powered capabilities that are related to three important busi of ARMC, which measures how quickly and easily a firm can respond to
ness needs: automating business processes, gaining insights, and opportunities within two areas: customer responsiveness and opera
engaging with customers (Davenport & Ronanki, 2018). tional adjustment. According to the underlying theory (e.g., Samba
These AI-powered capabilities include (1) AI-enabled automation murthy et al., 2003), the first-order dimensions of ARMC are
capability, also referred to as mechanical AI; (2) AI-enabled analytics complementary (i.e., they interact and co-vary with each other). Thus,
capability, known as cognitive insights or thinking AI; and (3) AI-enabled following guidelines provided by Tanriverdi and Venkatraman (2005),
relational capability, also known as feeling AI or cognitive engagement we conceptualize ARMC as a reflective, second-order construct con
(Davenport & Ronanki, 2018; Huang et al., 2019; Huang & Rust, 2021a). sisting of two first-order dimensions: customer responsiveness and
Although these three AI capabilities are interrelated, each supports operational adjustment. Furthermore, considering that tasks can differ
business functions differently (Huang & Rust, 2021b). Each AI-powered in their degree of automation, analytics, and relational intelligence
capability brings unique advantages: AI-enabled automation excels in capability (e.g., information acquisition often involves more mechanical
standardization, AI-enabled analytics capability is most effective for processes, whereas information processing and analysis are more
personalization, and AI-enabled relational capability proves ideal for analytics-driven) (Huang et al., 2019), we perceive the three AI-powered
fostering strong relationships (Huang & Rust, 2021a, 2021b). The defi capabilities as interrelated yet distinct constructs. Thus, we conceptu
nition of each AI-powered capability and examples of each are sum alize them as reflective, first-order constructs.
marized in Table 1. Furthermore, we consider the impact of environmental conditions as
The first AI capability—AI-enabled automation capability—is used to we explore the relationships between AI-powered capabilities and
automate repetitive and routine tasks (Huang & Rust, 2021b; Makarius ARMC. We hypothesize that the effects of AI-powered capabilities on
et al., 2020). AI-enabled automation is often used to improve output ARMC may vary depending on the environmental conditions. Specif
consistency, standardization, reliability, and precision (Huang et al., ically, we aim to comprehend the influence of perceived hostile and
2019). The second capability—AI-enabled analytics capability—can be dynamic environmental contexts (Mitchell et al., 2011). Finally, we
3
Fig. 1. Conceptual model.
establish a connection between ARMC and our primary outcomes of H3: AI-enabled automation capability is positively associated with
interest: firm performance, process innovation, and product innovation. AI-enabled relational capability.
In our investigation, we adopt an agility perspective to view firm per
formance as an outcome of ARMC (Chen et al., 2014; Holmqvist & Pessi, 3.2.2. AI-enabled automation capability and ARMC
2006; Kathuria et al., 2018). Additionally, we employ the dynamic ca AI-enabled automation capability plays a vital role in enhancing
pabilities perspective, which suggests that firms with strong sensing and ARMC by automating routine tasks and enabling team members to focus
adaptive response abilities can drive innovation and new product on more impactful high-level activities for the business (Raju & Koch,
development (Hoonsopon and Puriwat, 2019; Puriwat and Hoonsopon, 2019). A notable example is Crédit Agricole Technologies et Services
2021; Teece et al., 2016). Thus, we also measure process and product (CA–TS), which employs Robotic Process Automation (RPA) to auto
innovation as our dependent variables. mate its project validation process, resulting in faster decision-making
(Deloitte, 2020). AI-enabled automation not only breaks down organi
3.2. Hypotheses development zational siloes by automating data sensing and collection processes, but
also delivers standardized benefits through its consistency (e.g., using
3.2.1. The relationships among AI-Powered capabilities robots for packaging tasks) (Huang & Rust, 2021b). By eliminating un
According to Huang and Rust’s (2021a) perspective of multiple AIs, necessary traditional data collection methods, automation capability
the advancement of AI occurs cumulatively. In this context, AI pro helps firms respond to market changes by reducing redundancies and
gresses through higher levels while retaining its lower-level capabilities. inefficiencies (Huang et al., 2019).
From this perspective, AI-enabled automation represents the lowest and The strategic deployment of AI-enabled automation is also consid
simplest level, followed by AI-enabled analytics capability, and, finally, ered a contingency-driven decision, empowering organizations to
AI-enabled relational capability. Therefore, we assert that AI-enabled respond to market changes by automating routine and time-consuming
automation serves as the fundamental building block for firms’ ana processes, thereby reducing production costs and risks associated with
lytics and relational capabilities. AI-enabled automation empowers delivering products/services. Thus, AI-enabled automation can improve
firms to perform repetitive tasks, albeit with limited learning and scalability (Raju & Koch, 2019). In turn, firms can efficiently redeploy/
adaptive capabilities (Huang et al., 2019). Considering that this type of redirect their resources to value creating and value capturing activities
AI is the most cost-effective and easiest to implement (Davenport & (Teece et al., 2016). Therefore, we hypothesize the following:
Ronanki, 2018), we can reasonably assume that firms would typically H4: AI-enabled automation is positively associated with ARMC.
prioritize establishing this AI capability as their initial step. Once this
capability has been established, they can utilize it to enhance AI-enabled 3.2.3. AI-enabled analytics capability and ARMC
analytics capability through algorithms and models as a means of The relationship between AI-enabled analytics capability and ARMC
learning, as well as AI-enabled relational capability to establish mean can be explained through the lens of sensing (Teece et al., 2016). In
ingful relationships with users ((Huang & Rust, 2022)). environments characterized by uncertainty, companies must sense and
Further, AI-enabled analytics capability can significantly enhance generate digital options for growth before the market logic of these
firms’ relational capability, leading to improved business decision- options becomes apparent to all (Teece et al., 2016). AI functionalities,
making, especially in the context of future collaborations with human such as predictive analytics, diverse customer data analysis, and
users (Huang & Rust, 2022). This result becomes evident when AI- personalized content creation, provide a range of digital options for
powered analytics efficiently uncover valuable insights and person organizations. These options enhance process efficiency, knowledge
alize user interactions (Davenport et al., 2020). Furthermore, as AI alternatives, and rapid response to changes. For instance, from a market-
continually acquires new insights and learns through its algorithms, it access perspective, integrating AI into products and services allows
becomes increasingly adept at recognizing and responding appropri companies to generate and test models swiftly, enhancing the ability to
ately during human interactions (Huang et al., 2019). Thus, our first explore new markets, develop innovative products, and expedite prod
three hypotheses concern the relationships among AI-powered uct development (Davenport, 2018). From an operational-related
capabilities: standpoint, AI analytics capability offers novel problem-solving ap
H1: AI-enabled automation capability is positively associated with proaches and supports human decision-making by overcoming inherent
AI-enabled analytics capability. information-processing limitations (Haefner et al., 2021). Thus, AI-
H2: AI-enabled analytics capability is positively associated with AI- enabled analytics capability is crucial in facilitating ARMC, leading to
enabled relational capability. the personalization of products and services by supporting decision-
4
making processes throughout the entire value chain. business” (Zahra & Garvis, 2000, p. 475). These unfavorable environ
According to the dynamic capability perspective, undertaking ac mental conditions can result from radical industry changes, intense
tions to proactively create hypotheses about the future implications of regulatory burdens placed on the industry, or fierce rivalry among
observed events and trends can help organizations sense and seize competitors (Chen et al., 2014; Zahra & Garvis, 2000). Hostility also
market opportunities (Teece et al., 2016). For example, firms can use results from perceived competitive-, market-, and product-related un
machine learning to make predictions and assess their quality in practice certainties (Dess & Beard, 1984; Zahra & Garvis, 2000). Under hostile
(Albrecht et al., 2021; Capatina et al., 2020; Davenport et al., 2020). environmental conditions, market structures are fluid, clusters of know-
They can also track users’ actions and responses in the field, offering how and technologies are globally dispersed, and innovations are
accurate suggestions that capture users’ increasing interest over time dependent on technology combinations (Teece, 2014). Thus, environ
(Hollebeek et al., 2021). Since AI analytics tools provide data-driven mental hostility requires firms to break their routines and continuously
knowledge (Kiron & Schrage, 2019), they offer accurate and reliable learn new complexities and uncertainties, leading to increased chances
information, enabling firms to flexibly respond to market opportunities of survival and success in such challenging settings (Teece et al., 2016).
that might be overlooked by humans operating alone (Haefner et al., Not surprisingly, firms need a strategic direction to ensure survival in
2021). Thus, we hypothesize the following: such environments (Slevin & Covin, 1997).
H5: AI-enabled analytics capability is positively associated with Given that AI-enabled automation is primarily used to automate
ARMC. routine tasks (Huang et al., 2019), maximize efficiency, and minimize
variability (Huang & Rust, 2021a), firms relying on this capability while
3.2.4. AI-enabled relational capability and ARMC operating in highly hostile environments cannot quickly respond to
AI-enabled relational capability goes beyond merely detecting pat market changes. While AI-enabled automation enhances operational
terns in data and providing insights for decision-making processes; it efficiency and reduces human error (Huang & Rust, 2021a), its limita
involves using data to train AI systems that can interact with users tions lie in its inability to adapt quickly to dynamic and unpredictable
without human intervention (Davenport & Ronanki, 2018). AI’s rela market conditions (Tussyadiah, 2020). Thus,
tional capability is used to engage users in NLP chatbots, deep learning, H7a: Environmental hostility negatively moderates the relationship
and machine learning (Duan et al., 2019). When employed in user between AI-enabled automation capability and ARMC.
interaction, an AI application with a relational capability offers several In a hostile environment, accurate information is difficult to obtain
business benefits: it facilitates firm-initiated communications, engages (Bourgeois & Eisenhardt, 1988), making strategic decision-making
with customers in a friendly manner, is unaffected by frustration or fa difficult due to the uncertainty surrounding the potential significance
tigue like humans are, and can easily scale up to handle a large volume of certain changes (Özsomer et al., 1997). With AI-enabled analytics
of customer communications (Luo et al., 2019). These advantages capability, firms can anticipate and adapt to market trends (Allal-Chérif
empower firms to listen attentively to users and address their concerns et al., 2021). They can also create hypotheses about the future impli
effectively. cations of observed events and test these hypotheses to generate path
The relationship between relational capability and ARMC can also be ways for new products or services (Teece et al., 2016). Consequently, in
explained from the data network effects perspective (Gregory et al., a highly hostile environment, prioritizing AI-enabled analytics capa
2021). A platform exhibits data network effects if the more it learns from bility becomes imperative for achieving ARMC. Thus, we expect the
the data it collects on users, the more valuable the platform becomes to following:
each user. For example, Uber uses machine-learning algorithms to H7b: Environmental hostility positively moderates the relationship
analyze users’ data and improve its algorithmic matching, information, between AI-enabled analytics capability and ARMC.
and experience offered to users who engage in exchanges among riders Hostile environments are characterized by severe competitive in
and drivers. Uber’s goal is to inform and engage users with push noti tensity exacerbated by price wars and customers who exhibit little loy
fications and messages that provide intelligent recommendations that alty to their suppliers (Slevin & Covin, 1997). Among the three AI-
adapt to changing contextual and situational circumstances (Gregory powered capabilities, AI-enabled relational capability is the most com
et al., 2021; Rosenblat, 2018). One of the strategic roles of AI in today’s plex to develop, and many AI applications with this capability are still in
business is to continuously improve enterprises’ products and services the experimental stage (Huang et al., 2019). With its current progress in
for each user (Gregory et al., 2021), allowing firms to cope with rapid, the industry, developing AI with relational capability may subject firms
relentless, and uncertain changes. Therefore, we hypothesize the to higher risks in terms of resource allocation compared to developing AI
following: for analytics purposes. Given that hostile environments are unforgiving
H6: AI-enabled relational capability is positively associated with (Slevin & Covin, 1997), organizations operating in such conditions
ARMC. should leave little to chance and carefully consider the trade-offs
involved in investing resources involving AI with relational capabilities.
3.2.5. The moderating role of environmental hostility and environmental Additionally, in a hostile environment, the range of viable solutions
dynamism for any given organization will likely be restricted such that multiple
Prior research has shown that external environmental factors are paths will not be equally workable solutions (Slevin & Covin, 1997).
important contextual variables in influencing decision-making processes When AI with a relational capability is deployed in a hostile environ
and organizational agility (Chatterjee et al., 2023; Helfat, 2022; Lee ment, organizations put their relationship with their customers at risk
et al., 2015; Mitchell et al., 2011; Teece et al., 2016). These factors can since even the slightest mistake could drive them away, restricting firms’
significantly impact companies’ strategic choices, resource allocation, ability to remain flexible. For example, customers may feel uncomfort
and overall responsiveness to changing market conditions, thereby able talking to AI for personal needs or seeking chatbot assistance in
playing a crucial role in shaping organizations’ ability to adapt to and purchase decisions. Humans may also harbor prejudice, perceiving
thrive in a dynamic business landscape (Helfat et al., 2007). In this chatbots as lacking personal feelings and empathy, leading to doubts
study, we seek to understand the effect of two environmental contexts about the bots’ trustworthiness with personal information (Luo et al.,
(environmental hostility and environmental dynamism) in moderating 2019). Consequently, utilizing AI with a relational capability may
the relationship between AI-powered capabilities and ARMC. potentially decrease firms’ ability to respond to the market when they
operate in hostile environments. Thus, we hypothesize the following:
3.2.5.1. Hostile environmental context. Environmental hostility H7c: Environmental hostility negatively moderates the relationship
(Özsomer et al., 1997) indicates “unfavorable external forces for a firm’s between AI-enabled relational capability and ARMC.
5
3.2.5.2. Dynamic environmental context. A dynamic environment is contingency effects. These effects are observed as fluctuations in various
“one with a highly unpredictable and unstable rate of change and high components of the external environment, including customers, com
levels of uncertainty about the state of the context, the means-ends re petitors, and technology (Chakravarty et al., 2013). We have summa
lations, and/or the outcomes of actions” (Mitchell et al., 2011, pp. 687- rized the interaction effects of environmental conditions in Table 2.
688). These changes include customer and technology changes (Im &
Workman, 2004; Jaworski & Kohli, 1993; Kohli & Jaworski, 1990). For 3.2.6. ARMC and firm performance
example, the changes brought about by the driving force of technology We examine the relationship between ARMC and firm performance
have created hypercompetitive environments (Kyläheiko & Maijanen, by drawing upon the agility literature (e.g., Chen et al., 2014; Kathuria
2020; Teece, 2014). In such environments, customers’ demands and et al., 2018; Ngai et al., 2011; Queiroz et al., 2018; Ravichandran, 2018).
competitors’ actions are unpredictable, which requires frequent changes In this study, we define firm performance in terms of market and
in how businesses operate (Allal-Chérif et al., 2021). financial performance. ARMC can improve performance by expanding
Given that AI-enabled automation learns and adapts to only a min firms’ range of competitive actions and the nature of their feasible re
imal degree, it is ideal for service standardization (Huang & Rust, sponses to environmental change (Tallon & Pinsonneault, 2011). Firms
2021b). For instance, in a low level of dynamism where customers with high ARMC can take advantage of market opportunities to improve
exhibit homogeneous demand or when companies handle routine tasks, delivery performance, reduce reaction time (Anand & Ward, 2004), and,
AI-enabled automation offers a convenient and efficient operational in turn, achieve high performance (Chen et al., 2017). Firms with high
solution (Huang & Rust, 2021b). However, environmental dynamism ARMC, therefore, can expect future benefits in revenues or profitability,
requires firms to respond early and rapidly to market changes (Elazhary cost avoidance, and high market growth. Thus, we hypothesize the
et al., 2023; Lee et al., 2015). Under such dynamism, AI-enabled auto following:
mation will be a bottleneck. Its rigid and standardized processes may not H9: ARMC is positively associated with firm performance.
sufficiently cater to the diverse and fluctuating demands of customers or
adapt to unexpected shifts in the competitive landscape. Thus, when an 3.2.7. ARMC, process Innovation, and product innovation
environment is highly dynamic, overreliance on AI-enabled automation In organizational settings, innovation is commonly defined as
could lead to missed signals and inadequate decision-making, leaving creating any new product, service, or process for an organization
organizations vulnerable to market disruptions. Thus, (Tushman & Nadler, 1986). A vast majority of research supports that
H8a: Environmental dynamism negatively moderates the relation firm innovation is a multidimensional phenomenon, and one way to
ship between AI-enabled automation capability and ARMC. conceptualize innovation as an organizational outcome is to separate
AI-enabled analytics capability allows firms to create new informa product innovation from process innovation. Product innovation is “the
tion and predictions from data, even when the data are unstructured novelty and meaningfulness of new products introduced to the market
(Shrestha et al., 2019). This capability is akin to the generative-sensing [in] a timely fashion” (Wang & Ahmed, 2004, p. 304). Product inno
capability (Teece et al., 2016) that empowers firms to proactively sense vation is most often referred to as perceived newness, novelty, or
and learn from their environment. By harnessing AI-enabled analytics, uniqueness of products (Wang & Ahmed, 2004). It is also defined as any
organizations can gain valuable insights from vast and complex datasets, emerging technology or combination of emerging technologies that are
enabling them to identify emerging trends, anticipate market shifts, and exploited to produce goods or services (Bhoovaraghavan et al., 1996).
uncover patterns that might have otherwise remained undetected Process innovation is “the introduction of new production methods,
(Huang & Rust, 2021b; Teece et al., 2016). Possessing this capability management approach, and technology that can be used to improve
provides firms with flexibility and options that they can use to customize production and management processes” (Wang & Ahmed, 2004, p. 305).
and configure their resources (e.g., modify their market segments when Similarly, Trantopoulos et al. (2017) characterized process innovation
needed) as they operate in a dynamic environment (Teece et al., 2016). as a novel and efficiency-improving endeavor to reduce the production
Thus, environmental dynamism positively moderates the relationship costs of goods or services. In some studies, process innovation is
between AI-enabled analytics and ARMC because it amplifies the considered a sub-element of technological innovation (Wang & Ahmed,
importance of adaptability, flexibility, and data-driven insights within 2004). While product innovation is centered on the market and pri
organizations: marily driven by customer needs, process innovation is internally ori
H8b: Environmental dynamism positively moderates the relationship ented and primarily motivated by efficiency (Damanpour &
between AI-enabled analytics capability and ARMC. Gopalakrishnan, 2001). Regardless of the differences between product
With a high environmental dynamism, firms face significant unpre and process innovation, prior research has suggested that they are
dictability regarding their customers’ tastes and preferences (Lee et al., closely related (Oke et al., 2007).
2015). Thus, their IT initiatives should respond to emerging and fast- Recently, innovation has emerged as a key outcome of organizational
changing business needs (Chakravarthy et al., 2013). Unlike in a hos agility, as evidenced by studies conducted by Battleson et al. (2016),
tile environment where customers are unlikely to exhibit loyalty to Hoonsopon & Puriwat (2019), and Puriwat and Hoonsopon (2021).
firms, customers in a dynamic environment tend to seek adaptability, Considering this development, we contend that examining the impact of
responsiveness, and personalized experiences. In such an environment, ARMC on innovation becomes particularly relevant when AI-powered
AI-enabled relational capability may positively impact ARMC because it capabilities are involved. AI can assist the innovation process by
allows firms to nurture long-lasting relationships and foster customer improving conditions for innovation and, in turn, facilitating the idea-
loyalty (Huang & Rust, 2021a). The relational capability allows firms to generating process and overcoming information-processing constraints
continuously monitor and quickly improve product or service offerings (Haefner et al., 2021). For example, machine learning-based methods
to address customer needs (Lu & Ramamurthy, 2011). Relatedly, being are used to predict the most promising materials to test, providing firms
responsive to customers’ requests requires simultaneous adjustments to with the ability to adapt. The need to change and adapt triggers crea
firms’ internal business processes (Teece et al., 2016). Using AI to tivity and the ability to innovate (Georgsdottir & Getz, 2004). Thus, we
enhance customer relations can provide organizations with better op argue that AI indirectly influences firms’ innovation through ARMC.
portunities to respond to market demands and flexibly rearrange their The relationship between ARMC and innovation can be explained
internal business processes. Thus, from several perspectives. First, from the creativity perspective, the
H8c: Environmental dynamism positively moderates the relationship flexibility of the existing knowledge structure has proven to be a source
between AI-enabled relational capability and ARMC. of new ideas (Georgsdottir & Getz, 2004). Mumford and Gustafson
In conclusion, the impact of environmental factors on the relation (1988) also argued that the flexible use of existing knowledge plays a
ships between AI-powered capabilities and ARMC vary based on critical role in generating new ideas. Since innovation overlaps with
6
Table 2
Summary of the interaction effects of environmental conditions.
High AI-Enabled Automation High AI-Enabled Analytics High AI-Enabled Relational
High Env. Low ARMC High ARMC Low ARMC

Hostility AI-enabled automation is desirable in a stable AI-enabled analytics is desirable in a highly AI-enabled relational capability is desirable in a stable
environment (H7a) hostile environment (H7b) environment (H7c)
High Env. Low ARMC High ARMC High ARMC
Dynamism AI-enabled automation is desirable in a stable AI-enabled analytics is desirable in a highly AI-enabled relational capability is desirable in a highly
environment (H8a) dynamic environment (H8b) dynamic environment (H8c)
creativity, we infer that ARMC is related to process innovation. Second, enabled automation, AI-enabled analytics, and AI-enabled relational capa
prior organizational research has shown that when organizations are bilities. Consistent with the literature (e.g., Huang et al., 2019; Huang &
rigid and thus agility is low, company managers prefer conservative Rust, 2021a), we modeled each dimension as a first-order construct.
decisions, avoid risky behaviors, and consequently inhibit the process of Then we developed a new scale to operationalize AI capabilities based
innovation (Hisrich, 1990). Further, in a meta-analysis of organizational on insights from Cheatham et al. (2020) and used three items to measure
innovation, Damanpour (1991) found that among other factors, a pos AI-enabled automation. All three items measured the extent to which AI
itive managerial attitude toward change and the existence of special tools and applications are used for automation. For example, AI tools
units that focus on change and adaptation processes facilitate innova and applications are used to automate time-consuming activities,
tion. In the context of technology capability, Zhou and Wu (2010) noted including claim processing, basic customer-service interaction, and in
that “as a firm accumulates more technological capability, its absorptive ventory tracking. We measured AI-enabled analytics capability using
capacity should increase, which leads to more product innovations” (p. nine items, including the use of AI to predict equipment failure, respond
549). Firms with high ARMC are better at reacting to changes in market quickly to changing inventory levels, test nearly all potential scenarios
demand (Tallon & Pinsonneault, 2011). They can transform resources before making a decision, predict the next-best product or service a
and capabilities to stimulate the development of new products (Hoon customer is likely to buy, and simulate product or service delivery. To
sopon & Puriwat, 2019). Thus, we hypothesize the following: measure AI-enabled relational capability, we used four items that reflect
H10: ARMC is positively associated with (a) process innovation, and the extent to which firms used AI to engage with their customers. These
(b) product innovation. items were adapted from Lee et al. (2020).
To validate all items used in measuring AI-powered capabilities, we
4. Research methodology asked for feedback on our initial measurement items from two IS faculty
members with experience in survey research methods. The feedback led
This study uses a two-stage survey to collect data from IT executives us to review and modify several items. Next, the initial set of items and
and business decision-makers (e.g., C-level executives, presidents, senior the predefined categories were submitted for a card-sorting test (Moore
vice presidents, senior directors, directors, and managers or senior & Benbasat, 1991). Results from this test indicated that wording for
managers). These key informants are likely to be most informed about some items needed additional clarity. After we conducted two rounds of
their firms’ business and AI initiatives. We measured AI constructs, card-sorting tests, the test resulted in a satisfactory classification of items
environmental factors, ARMC, and all control variables at one point in into predefined categories.
time (t). Then we measured firm performance, process innovation, and ARMC: We measured ARMC using seven items that reflected how
product innovation at the second point in time (t + 1). The first survey easily and quickly firms can address the opportunities in two areas:
was conducted in November 2020, and the second was completed by the customer responsiveness and operational adjustment. These items were
same respondents four months later. We chose the four-month gap to adapted from Tallon and Pinsonneault (2011).
minimize response bias and ensure the accuracy of the information Environmental Hostility and Dynamism: We measured environ
collected about the firms’ performance. Conducting the second survey mental hostility using five items developed by Chen et al. (2014). The
early in 2021 allowed us to capture up-to-date data and insights about indicators assessed the existence of unfavorable external forces in firms’
firms’ performance in the previous year. business environments (Zahra & Garvis, 2000). Environmental dyna
We used a longitudinal study design primarily for three reasons: (1) mism assessed the degree of rapid technological change, shifts in cus
to achieve correspondence with the temporal ordering of constructs in tomers’ preferences, and changes in competitor actions. The scale was
our research model; (2) to enhance our arguments for causality adapted from Lee et al. (2015).
regarding the relationship between agility and organizational outcomes; Firm Performance: We used self-reported items to measure firm
and (3) to reduce the threats of common method bias. Data were performance. This approach is consistent with prior IT capability
collected through an online channel with organizations as our unit of research investigating relative performance measures (e.g., Chen et al.,
analysis. Next, we discussed our methodological choices, encompassing 2014; Queiroz et al., 2018; Ravichandran et al., 2005). We adapted five
concerns regarding measurement selection, sampling, response rate, and items from Queiroz et al. (2018) to measure market share, revenues,
evaluation of potential non-response bias, following the recommenda sales growth, and profitability relative to competitors.
tions of Hulland et al. (2018). Process and Product Innovation: We modeled firms’ product and
process innovation as reflective, first-order constructs (Damanpour &
Gopalakrishnan, 2001). Product innovation captured firms’ capacity to
4.1. Measures innovate compared to all other competitors within the same industry
concerning the level of novelty, utilization of cutting-edge technology,
To the extent possible, we operationalized constructs by adapting pace of product development, and number of new products or services
existing scales to the context of our current studies. We captured re introduced. On the other hand, process innovation denotes firms’
sponses to all items using multi-item reflective measures on a seven- technological competitiveness, speed in adopting the latest technology,
point Likert scale. We then validated the measurement scales in a se and novelty in technology utilization. These items were adapted from
ries of procedures to ensure content validity, construct validity, and Prajogo and Ahmed (2006).
reliability (Straub, 1989), as described later in this section. The details of Control Variables: We controlled for the effect of organizational
measurement items can be found in Appendix A. factors (i.e., firm location, industry sector, firm size, and firm age) on
AI-Powered Capabilities: We measured three AI capabilities: AI-
7
firm performance and innovation following prior work on the business Table 3
value of IT (see Chen et al., 2021; Kulkarni et al., 2017). Firm’s location Demographic respondents.
was a binary variable with 0 for UK companies and 1 for French com Demographic Criteria Total
panies. Industry sector was a binary variable with 0 for non- Information (%)
manufacturing firms and 1 for manufacturing firms. Firm size was Firm’s Geographic France 73
measured as a firm-wide number of employees, and firm age was Location (68.2)
measured as the number of years a firm has been in business. United Kingdom 34
(31.8)
Industry Type Banking 8 (7.5)
4.2. Survey development and administration Computer/software 32
(29.9)
We pilot tested the final questionnaire with Prolific.co using the UK Consulting 10 (9.3)
sample. Two screening criteria were used for the pilot study: (1) Subjects Insurance 5 (4.7)
Manufacturing 14
had supervisory responsibilities at their companies, and (2) their firms
(13.1)
had adopted AI applications in recent years. After receiving 94 valid and Medicine/health 7 (6.5)
complete responses in our pilot study, we reviewed and refined the items Publishing/communications 8 (7.5)
based on the results from the pilot study. Hotel/restaurants 5 (4.7)
For the actual data collection, we asked a European professional Transportation 6 (5.6)
Energy 4 (3.7)
market research company to contact IT and business decision-makers in Other (constructions, trading and retail, 8 (7.5)
diverse organizations. Our population of interest included firms that commerce, education)
have adopted AI in Europe (i.e., France and the United Kingdom). For Company Age < 5 years 2 (1.9)
data collection from France, survey items were prepared in English and 5–10 years 12
(12.1)
then translated into French. Despite the cultural differences, France and
6–15 years 22
the UK share several commonalities, notably that both countries have (20.6)
stable economies and strong jurisdictions in Europe. They are also > 15 years 70
characterized by the same values, such as high autonomy and individual (65.4)
financial security, with a strong emphasis on achievement and individ Number of Employees <50 15 (14)
50–200 15 (14)
ual initiative (Mellahi, 2001). According to OECD statistics (Organiza 201–500 16 (15)
tion for Economic Co-operation & Development, 2019), the rates of 501–1000 17
technology adoption in France and the UK were 90.17 and 95.85, (15.9)
respectively. Therefore, both countries are comparable in terms of 1001–2500 15 (14)
2501–10,000 14
technology adoption and business practices.
(13.1)
The samples for both countries were drawn from the panel members >10,000 14 (14)
of the market research firm. We utilized the prescreening filters pro Respondent Position C-Suite Level Executive 17
vided by this company to select appropriate respondents, such as IT or (15.9)
business executives. Additionally, we included two filter questions in Vice President 2 (1.9)
General Manager 16 (15)
our survey, measured on a five-point Likert scale, to ascertain whether Business Unit Director 11
respondents had sufficient knowledge about their companies’ business (10.3)
and IT strategies. Only those who indicated adequate knowledge (above IT Director 23
the cut-off point of three out of five) were eligible to participate in the (21.5)
Business Manager 20
study. The average IT knowledge of respondents was 3.88 (SD = 0.87),
(18.7)
while their average business knowledge was 3.87 (SD = 0.86). These IT Manager or Project Manager 18
figures demonstrate a relatively balanced and comparable level of (16.8)
expertise in both areas. Four months after completing the first survey,
the same respondents were contacted to complete the second survey. Of
225 respondents who completed the first survey, a total of 107 re sample (t-value = -1.98; p =.05).
spondents completed the second survey for an effective response rate of
47.5 percent. We used two mechanisms to ensure that the same indi 5. Data analysis and results
vidual responded to both surveys for the same organization. First, each
panelist had a unique identification number, allowing us to match the 5.1. Measurement validation
two data sets. Second, we asked several personal questions and matched
their responses in both surveys. The matching technique provides sup We used Partial Least Squares (PLS) in the package WarpPLS 7.0
port that the same individual responded to both surveys. (Kock, 2022) to perform the data analysis. PLS is appropriate for the
study because it handles second-order constructs better than covariance-
4.3. Sample characteristics based structural equation modeling (CB-SEM) does. It also assesses a
measurement model for the theoretical model, makes no assumptions
Data were collected from firms in France (68.2 %) and the U.K. (31.8 regarding data normality, and can handle small samples (Kathuria et al.,
%). We provided a summary description of the sample in Table 3. A 2018). The first stage of data analysis focused on the measurement
comparison of earlier with later respondents from each country group properties of constructs. We assessed each construct’s convergent and
regarding the firm’s age and size indicated no differences, reducing discriminant validity by factor-analyzing items grouped under their
concerns of non-response bias (Armstrong & Overton, 1977). Similar respective construct and the item-total correlations for all constructs.
comparisons were made among participants from the two countries. The Several items were dropped due to a wording issue and/or low loadings.
analysis indicated that the two countries’ respondents were statistically After we dropped this item, all remaining items loaded highly on the pre-
similar regarding key study variables and firms’ characteristics. The specified construct (see Appendix B).
only variable that showed a slightly significant difference between the Table 4 shows the descriptive statistics of all the constructs,
two countries was environmental hostility, with the UK sample report including means, standard deviations, and latent variable coefficients.
ing a slightly higher score of environmental hostility than the France The average variance extracted (AVE) values for all reflective constructs
8
Table 4 variables on the dependent variables, and Models 2b, 3b, and 4b explore
Descriptive statistics and latent variable coefficients. the effects of ARMC on the dependent variables.
Construct Mean SD CR AVE As illustrated in Fig. 2, AI-enabled automation was positively asso
ciated with AI-enabled analytics and AI-enabled relational capabilities,
AI-enabled automation capability 5.28 1.14 0.88 0.72
AI-enabled analytics capability 4.94 1.34 0.96 0.74 and AI-enabled analytics capability was positively associated with AI-
AI-enabled relation capability 5.28 1.15 0.93 0.76 enabled relational capability. Thus, H1, H2, and H3 were supported.
Adaptive response to market change (ARMC) 5.35 0.88 0.92 0.54 In Model 1a, the direct effects of AI-enabled automation, AI-enabled
Environmental hostility 4.34 1.61 0.95 0.80 analytics capability, and AI-enabled relational capability on ARMC
Environmental dynamism 5.31 1.14 0.87 0.78
Process innovation 5.03 1.07 0.94 0.83
were significant. However, adding the interaction terms to the model
Product innovation 4.96 107 0.93 0.75 significantly altered the direct effects of AI-powered capabilities.3
Firm performance 4.99 1.19 0.95 0.79 Hence, we used Model 1b as our final model. The effect of AI-enabled
Respondent IT knowledge 3.88 0.87 1.00 1.00 automation and AI-enabled analytics on ARMC remained significant
Respondent business knowledge 3.87 0.86 1.00 1.00
even after the interaction terms were added to the model, supporting H4
CR = Composite Reliability; AVE = Average Variances Extracted; Firm IT and H5. However, the effect of AI-enabled relational capability became
knowledge and business knowledge were measured using a 5-point Likert Scale. insignificant, failing to support H6.
As expected, environmental hostility negatively moderated the
were greater than the minimum recommended value of 0.50, indicating relationship between AI-enabled automation and ARMC, supporting
that the items satisfied the convergent validity. Further, to confirm the H7a. Environmental hostility positively moderated the relationship be
scale reliability and internal consistency of the constructs, we calculated tween AI-enabled analytics capability and organizational agility, sup
the composite reliability scores (Fornell & Larcker, 1981). A composite porting H7b. As we hypothesized, the interaction between AI-enabled
reliability value of 0.70 or greater is considered acceptable. As reported relational capability and environmental hostility was negative and sig
in Table 4, the composite reliability scores for all the constructs were nificant. Thus, H7c was supported. However, the interactions between
greater than 0.88, demonstrating that all constructs had adequate reli AI-enabled automation and environmental dynamism, and between AI-
ability assessment scores. To ensure the discriminant validity of con enabled analytics and environmental dynamism, were not significant.
structs in the research model, we compared the square root of the AVE Thus, H8a and H8b were not supported. The interaction between AI-
for each construct with the other correlation scores in the correlation enabled relational and environmental dynamism was significant, sup
matrix (see Table 5). As shown in Table 5, all AVE values’ square roots porting H8c. When the interaction effects of environmental conditions
were greater than the correlations between the respective constructs and were included, the adjusted R2 of ARMC was significantly increased by
other latent constructs (Fornell & Larcker, 1981). seven percent (p <.001), indicating a medium-to-large-effect size (f2 =
0.22).4 For better visual clarity, these interaction effects are presented as
5.2. Test of common method bias a 3D graph in Appendix C (Fig. C1).
Further, ARMC was positively associated with firm performance,
In addition to using a longitudinal study design, we performed a supporting H9. Consistent with H10, ARMC had a significant effect on
series of statistical tests to alleviate common method concerns. First, we both process and product innovation. To further validate the indirect
applied Harman’s single-factor test using exploratory factor analysis effects of ARMC, we tested the significance of both indirect and
(Podsakoff et al., 2003). The resulting unrotated component matrix moderated indirect effects (if the relationship consists of both mediation
showed that the measurement items did not load on a single factor. A and moderator). As shown in Table 7, 10 out of 18 conditional indirect
single factor accounted for only 38.02 % of the total variance. Second, effects of AI-powered capabilities on organizational outcomes were
we examined common method variance (CMV) using a marker variable significant, suggesting meaningful moderating roles of environmental
technique (Lindell & Whitney, 2001) for the variables collected at one conditions.
point in time (t). In this technique, a marker variable is a scale that is
theoretically unrelated to at least one other scale in the questionnaire.
The marker variable’s mean correlation with other variables in the study 5.4. Test of endogeneity
(rM) is an estimate of the extent of CMV in the sample. We computed
CMV-adjusted correlations between the variables of interest (rA) by When causal effects are examined, endogeneity issues may arise from
partialling out rM from the uncorrected bivariate correlation (rU) using omitted variable bias. To address the potential endogeneity issues, we
the following formula rA = (rU - rM) / (1- rM) (Malhotra et al., 2006) and adopted an instrumental variable approach introduced by Kock (2022)
we estimated the significant difference between rU and rM using a t- in WarpPLS. A valid instrumental variable should be uncorrelated with
statistic (Bhattacherjee & Hikmet, 2007). In our study, we used re the error (i.e., satisfy the exclusion restriction) and correlated with the
spondents’ current state of mind (i.e., “my mind feels unfocused right endogenous regressor (i.e., satisfy the relevance condition) (Tan et al.,
now”) as the marker variable. This variable had a mean absolute cor 2017). In other words, the instrument should affect the dependent
relation of 0.07 with the items used in this study. The mean change in variables (i.e., firm performance and innovation) only through the
correlation due to the CMV-adjusted correlations was very small (0.03). endogenous regressor (ARMC). We proposed IT infrastructure compat
These results suggest the lack of CMV bias in our sample. ibility as a valid instrumental variable (i.e., it highly correlates with
ARMC (r = 0.58, p <.001) but does not correlate with the organizational
5.3. Structural model
3
We used WarpPLS 7.0 to assess the structural model. Since ARMC is The presence of the interaction term alters the strength or direction of the
conceptualized as a second-order aggregate construct, we generated relationships between the independent variables and the dependent variable,
rendering the main effects non-significant. This result suggests that the re
latent variable scores for each of its first-order dimensions, which were
lationships between the independent variables and dependent variables vary
then used as reflective measures of the second-order aggregate con
across different levels of moderators. Thus, the relationships should be inter
structs (Polites & Karahanna, 2012). Fig. 2 displays the final result of the preted as “conditional” effects at the different values of the included moderator
structural model, and Table 6 presents the results of hierarchical in the model, and the main effects become less relevant (Frazier et al., 2004).
regression analyses. Model 1a specifies the direct effects of AI-powered 4
Effect size (ƒ2) is calculated by the formula (R2full – R2partial)/(1 – R2full)
capabilities on the dependent variables. Model 1b tests the interaction (Mathieson et al. 2001). Cohen (1988) suggested 0.02, 0.15, and 0.35 as
effects model. Models 2a, 3a, and 4a specify the effects of control operational definitions of small-, medium-, and large-effect sizes, respectively.
9
Table 5
Correlation matrix of the focal constructs.
Construct 1 2 3 4 5 6 7 8 9 10 11 12 13
1 AI-enabled automation 0.85

2 AI-enabled analytics 0.71*** 0.86
3 AI-enabled relation 0.66*** 0.62*** 0.87
4 ARMC 0.70*** 0.61*** 0.62*** 0.92
5 Env. Hostility 0.31** 0.46*** 0.20* 0.26** 0.90
6 Env. Dynamism 0.50*** 0.67*** 0.55*** 0.67*** 0.21* 0.88
7 Firm performance 0.24* 0.40*** 0.31** 0.40*** 0.20* 0.27** 0.89
8 Process innovation 0.31** 0.35*** 0.22* 0.31** 0.24* 0.19 0.58*** 0.91
9 Product innovation 0.32*** 0.35*** 0.31** 0.35*** 0.24* 0.25** 0.64*** 0.77*** 0.87
10 Industry type -0.22* -0.12 -0.25* -0.22* 0.04 -0.11 -0.15 -0.22* -0.24* –
11 Firm age -0.05 -0.19* -0.01 -0.06 -0.27** -0.08 -0.02 -0.12 -0.09 0.00 –
12 Firm size -0.04 -0.06 0.11 0.07 -0.15 -0.01 0.11 0.01 0.02 -0.23* 0.26** –
13 Firm location 0.01 -0.14 0.05 -0.01 -0.19 -0.06 -0.05 -0.03 -0.04 -0.16 0.36*** 0.17 –
Notes: Square roots of average variances extracted (AVEs) are shown on the diagonal; ***p <.001; **p <.01; *p <.05.
Fig. 2. Structural model.
Table 6
Statistical results (standardized path estimates).
DV: ARMC DV: Firm Performance DV: Process Innovation DV: Product Innovation
Model 1a Model 1b Model 2a Model 2b Model 3a Model 3b Model 4a Model 4b
Adj. R2 0.53 0.68 0.00 0.13 0.03 0.09 0.04 0.13

Direct Effects
AI-enabled automation 0.42*** 0.34***
AI-enabled analytics capability 0.16* 0.14†
AI-enabled relational capability 0.24** 0.10 (ns)
ARMC 0.38*** 0.27** 0.30***
Env. Hostility 0.06 (ns)
Env. Dynamism 0.43***
Moderating Effects
AI-enabled auto x Env. Hostility -0.23**
AI-enabled analytics x Env. Hostility 0.54***
AI-enabled relational x Env. Hostility -0.32***
AI-enabled auto x Env. Dynamism -0.01 (ns)
AI-enabled analytics x Env. Dynamism -0.08 (ns)
AI-enabled relational x Env. Dynamism 0.35**
Control Variables
Firm location 0.08 -0.07 (ns) -0.03 (ns) -0.02 (ns) -0.04 (ns) -0.05 (ns)
Industry type -0.14† -0.06 (ns) -0.23** -0.17* -0.26** -0.19*
Firm size 0.09 0.08 (ns) -0.02 (ns) -0.02 (ns) -0.01 (ns) -0.02 (ns)
Firm age -0.02 (ns) 0.01 (ns) -0.10 (ns) -0.09 (ns) -0.09 (ns) -0.05 (ns)
***p <.001; **p <.01; *p <.05; †p <.10; The adjusted R2 of the model with the direct effects of AI-powered capabilities and environmental conditions was 0.61 (for
brevity, this model is not included in the table).
outcomes). The results show that the instrumental variable was insig assess whether endogeneity was an issue. First, we regressed ARMC on
nificant, suggesting that endogeneity is not a major issue in our study. all exogenous and control variables, including IT infrastructure
Further, we conducted the Durbin-Wu-Hausman test in Stata to compatibility. Then, we used the residual of the regression as an
10
Table 7 Table 7 (continued )

Analysis of indirect and conditional indirect effects.5 Relationship (Conditional) Type of
Relationship (Conditional) Type of Indirect Effects (p- Mediation/
Indirect Effects (p- Mediation/ value) Indirect Effects
value) Indirect Effects
AI-enabled automation*environmental -0.07 No moderated
AI-enabled automation → AI-enabled 0.22*** Partial mediation hostility → ARMC → product mediation
analytics → AI-enabled relational innovation
AI-enabled automation → AI-enabled 0.10* Partial mediation AI-enabled analytics*environmental 0.17** Moderated
analytics → ARMC hostility → ARMC → product partial mediation
AI-enabled automation → AI-enabled 0.04 No mediation innovation
relational → ARMC AI-enabled relational*environmental -0.10† Indirect only,
AI-enabled analytics → AI-enabled 0.03 No mediation hostility → ARMC → product moderated full
relational → ARMC innovation mediation
AI-enabled automation → ARMC → 0.14* Indirect only, full Environmental hostility → ARMC → 0.02 No mediation
firm performance mediation product innovation
AI-enabled analytics → ARMC → firm 0.05 No mediation AI-enabled automation*environmental 0.00 No moderated
performance dynamism → ARMC → product mediation
AI-enabled relational → ARMC → firm 0.04 No mediation innovation
performance AI-enabled analytics*environmental -0.04 No moderated
AI-enabled automation*environmental -0.09 †
Indirect only, dynamism → ARMC → product mediation
hostility → ARMC → firm moderated full innovation
performance mediation AI-enabled relational*environmental 0.11* Moderated
AI-enabled analytics*environmental 0.21*** Moderated dynamism → ARMC → product partial mediation
hostility → ARMC → firm partial mediation innovation
performance Environmental dynamism → ARMC → 0.13* Partial mediation
AI-enabled relational*environmental -0.12* Indirect only, product innovation
hostility → ARMC → firm moderated full
Note: Only indirect effects for paths with two segments were reported; ***p
performance mediation
Environmental hostility → ARMC → 0.03 No mediation <.001; **p <.01; *p <.05; †p <.10.
5
firm performance We also measured the direct effects of AI-enabled automation capability, AI-
AI-enabled automation*environmental 0.01 No moderated enabled analytics capability, and AI-enabled relational capability on firm per
dynamism → ARMC → firm mediation formance, process innovation, and product innovation. The only significant
performance capability was AI-enabled analytics capability to all three dependent variables,
AI-enabled analytics*environmental -0.05 No moderated with the path coefficients as follows: AI-enabled analytics to firm performance
dynamism → ARMC → firm mediation = 0.44, p < 001; AI-enabled analytics to process innovation = 0.29; p <.001; and
performance
AI-enabled analytics to product innovation = 0.28; p <.001. Adding ARMC and
AI-enabled relational*environmental 0.13* Indirect only,
dynamism → ARMC → firm moderated full
interaction terms to the model reduced these direct effects to 0.27, 0.27, and
performance mediation 0.26, respectively (all are significant at p <.01).
Environmental dynamism → ARMC → 0.16** Partial mediation
firm performance additional regressor in our hypothesized equations. The parameter es
AI-enabled automation → ARMC → 0.10† Indirect only, full
timate for the residual was not significant, indicating that ARMC was not
process innovation mediation
AI-enabled analytics → ARMC → 0.04 No mediation endogenous in our study. Further, the Durbin (score) (X2 = 0.78; p =.38)
process innovation and the Wu-Hausman test were not significant (p =.40). These insig
AI-enabled relational → ARMC → 0.03 No mediation nificant results indicate that ARMC was exogenous.
process innovation
AI-enabled automation*environmental -0.06 No moderated
hostility → ARMC → process mediation 6. Discussion
innovation
AI-enabled analytics*environmental 0.14* Moderated The primary objective of this study is to explore how AI-powered
hostility → ARMC → process partial mediation capabilities can add value to organizations by enhancing their ARMC.
innovation
We have proposed that three AI-powered capabilities—AI-enabled
AI-enabled relational*environmental -0.09† Indirect only,
hostility → ARMC → process moderated full automation, AI-enabled analytics, and AI-enabled relational capa
innovation mediation bilities—play pivotal roles in driving ARMC. Drawing on the perspec
Environmental hostility → ARMC → 0.02 No mediation tives of organizational agility and the dynamic capability framework, we
process innovation
conceptualize ARMC as firms’ capacity to swiftly recognize and adapt to
AI-enabled automation*environmental 0.00 No moderated
dynamism → ARMC → process mediation
market shifts, encompassing customer responsiveness and operational
innovation adjustment as their core competencies. We also examine the moderating
AI-enabled analytics*environmental -0.04 No moderated effects of environmental conditions (environmental hostility and dyna
dynamism → ARMC → process mediation mism) on the relationships between AI-powered capabilities and ARMC.
innovation
Our findings reveal that the relationships between AI-powered ca
AI-enabled relational*environmental 0.09† Indirect only,
dynamism → ARMC → process moderated full pabilities and ARMC depend on environmental conditions. Specifically,
innovation mediation the findings show that environmental hostility negatively moderates the
Environmental dynamism → ARMC → 0.11* Partial mediation relationship between AI-enabled automation and ARMC, as well as be
process innovation
tween AI-enabled relational capability and ARMC. However, it posi
AI-enabled automation → ARMC → 0.11* Indirect only, full
product innovation mediation
tively moderates the relationship between AI-enabled analytics and
AI-enabled analytics → ARMC → 0.04 No mediation ARMC. Additionally, the findings demonstrate that environmental
product innovation dynamism positively moderates the relationship between AI-enabled
AI-enabled relational → ARMC → 0.03 No mediation relational capability and ARMC. Finally, our study establishes signifi
product innovation
cant positive relationships between ARMC and both firm performance
and innovation. These findings have important implications for both
research and practice.
11
6.1. Implications for research which, in turn, facilitates the development of ARMC in a highly dynamic
environment. By leveraging AI-enabled relational capability, organiza
Our study makes several contributions to research on the business tions can adapt their customer interactions and responses in real time,
value of IT. First, the results provide robust empirical evidence sup enabling them to stay responsive to the ever-changing market conditions
porting the idea that AI can enhance organizational responsiveness and and better meet customer needs and preferences.
adaptability in the face of market changes. The hype around AI is Lastly, our findings strongly support a positive relationship between
causing many companies to prematurely adopt AI applications without ARMC and firm performance, process innovation, and product innova
fully grasping the true value of AI (Brock & von Wangenheim, 2019; tion. From the dynamic capabilities perspective, ARMC empowers firms
Davenport, 2018). However, directing AI toward the key business to seize new opportunities (Teece et al., 2016). However, being highly
competencies can strengthen firms’ ability to be adaptive, agile, and responsive and adaptive to the environment can come with costs.
responsive to market changes, thereby positioning firms for sustainable Consequently, the necessity for ARMC, and thus organizational agility,
growth and success in dynamic business environments. arises when substantial uncertainty and dynamism are present in the
Second, our findings suggest a positive association between AI- business environment (Chakravarty et al., 2013; Teece et al., 2016).
enabled automation and ARMC. By leveraging AI-enabled automation Moreover, AI can potentially revolutionize the way firms operate,
tools, such as RPA, firms can automate repetitive and time-consuming transforming an entire organization’s rigid mindset into a more
business processes, resulting in improved productivity, faster market responsive and adaptive approach. This transformation provides an
entry for products and services, and enhanced responsiveness to market opportunity to explore problems from multiple perspectives, paving the
opportunities. However, our findings indicate that AI-enabled automa way for innovative and novel solutions.
tion is more suitable for stable environments with well-defined pro
cesses and relatively predictable changes, where firms can produce and 6.2. Practical and societal implications
sell homogenous products in large volumes (Lee et al., 2015). Thus, AI-
enabled automation efficiently streamlines production and operational Our findings provide several practical implications. First, executives
processes, ensuring consistency and cost-effectiveness while minimizing exploring the potential of AI can use our results to guide their AI in
human errors and improving overall efficiency. vestments. Our study shows that different functions of AI can enhance
Third, our findings also reveal that the effect of AI-enabled analytics ARMC, depending on the environmental conditions. With the uncer
on ARMC is significantly altered once the moderating effect of envi tainty surrounding AI, executives would be wise to systematically
ronmental hostility is added to the model. Specifically, in highly hostile evaluate their business needs to determine which areas could benefit
environments where uncertainties and disruptions are prevalent, the from AI. The three AI-powered capabilities discussed in this study may
impact of AI-enabled analytics on ARMC becomes even more pro help organizations to broaden their digital options and flexibly respond
nounced. AI-enabled analytics can assist organizations in processing vast to market opportunities.
amounts of complex data and extracting actionable insights, allowing Our findings suggest that not all AI capabilities hold the same sig
them to gain a deeper understanding of market dynamics and customer nificance. AI-enabled automation can aid organizations in cost reduction
behavior (Huang et al., 2019). In such hostile conditions, the ability to and standardizing routine tasks, particularly in stable environments. On
quickly identify emerging trends and adapt business strategies accord the other hand, AI-enabled analytics capability may enhance firms’
ingly becomes a critical advantage for firms seeking to control the ARMC by processing, evaluating, and analyzing information, proving
inherent risk. Insights enabled by AI overcome the challenges of more valuable in a hostile environment. Lastly, AI-enabled relational
obtaining information in a hostile environment, enabling evidence- capability is key in fostering ARMC by actively engaging with customers
based decisions in complex situations (Fernández-Rovira et al., 2021). and continuously learning from these interactions, making the process
Thus, AI-enabled analytics capability allows firms to develop planned instrumental for adaptability in highly dynamic environments. By
strategies in response to constant threats posed by the environment. leveraging the diverse capabilities of AI, organizations can effectively
Fourth, our results demonstrate that environmental hostility nega adapt to changing environments and improve their performance.
tively moderates the relationship between AI-enabled relational capa Our findings also suggest that nurturing their capacity for swift and
bility and ARMC. AI tools with relational capability utilize existing effective responses to market shift enables organizations to seize new
conversation data to understand a wide range of customers and employ opportunities. This study highlights the positive impact of ARMC on
machine learning and NLP to grasp the context, continuously enhancing innovation outcomes. Although implementing AI can be challenging, AI
AI’s responses and interactions with customers (Davenport & Ronanki, systems can drive ARMC, leading to both process and product innova
2018). These diverse digital options enabled by AI relational capability tion, as well as improved performance outcomes. Therefore, AI systems
can potentially drive ARMC both at market and operational levels as enable firms to explore novel approaches to problem-solving, providing
organizations become more effective in responding to customers, lead them with the flexibility to respond to threats and opportunities
ing to changes in internal business processes. However, many AI systems (Haefner et al., 2021). With such flexibility, firms can readily identify
designed for this purpose are still experimental (Huang & Rust, 2020). emerging market demands and create new products/services to meet
Thus, in highly hostile environments where the risk is too high and can those needs.
be costly (Mitchell et al., 2011), the effectiveness of AI-enabled rela Our study indicates that AI capabilities are transforming the dy
tional capability may be hampered, and organizations should carefully namics between humans and machines, the methods of data analysis,
evaluate its implementation and consider alternative strategies to and the decision-making process from a societal standpoint. Although AI
enhance ARMC under such conditions. initiatives can greatly influence customer experience and workforce
Fifth, we found that environmental dynamism does not moderate the skills (Butner et al., 2020), how these changes will unfold in the long run
relationship between AI-enabled automation and ARMC, as well as be is currently unknown. For instance, past research has shown that under
tween AI-enabled analytics and ARMC. However, we observed a positive high levels of environmental hostility and dynamism, IT resources and
and significant moderation effect of environmental dynamism on the competencies are necessary so that managers can actively sense op
relationship between AI-enabled relational capability and ARMC. In portunities for competitive actions (Lee et al., 2015). AI has transformed
environments characterized by high dynamism, firms often face signif how organizations perceive and respond to such challenges since
icant unpredictability regarding their customers’ tastes and preferences AI-powered capabilities provide a more flexible and adaptive approach.
(Lee et al., 2015). Consequently, they need to continuously engage in As AI technologies continue to advance, they are likely to play an even
competitive actions to attract and retain customers. In this context, AI- more significant role in shaping the future of business and society.
enabled analytics can aid in establishing a relational capability, However, alongside the tremendous potential are ethical, regulatory,
12
and privacy concerns that need to be carefully addressed to ensure method bias does not appear to threaten the validity of our results sta
responsible and sustainable AI deployment (Whittlestone et al., 2019). tistically, using the same respondent might have upwardly biased our
Thus, continued research on the impacts of AI and its potential chal results. Thus, we recommend that future studies consider research de
lenges is essential to harnessing its full benefits while mitigating po signs that allow data collection from multiple respondents within an
tential risks. organization. Future research could also complement the self-reported
measures of performance with more objective performance data.
6.3. Limitations and future research directions Lastly, our data were collected from firms located in France and the U.K.,
and our sample includes a broad spectrum of industry groups. Although
This research has some limitations that suggest directions for future we controlled for the effect of location and industry types, investigating
work. First, as mentioned earlier, organizational agility is a compre data from a more homogeneous population could result in different
hensive and multifaceted concept. While we derived the concept of conclusions.
ARMC from the broader agility framework, we encourage future
research to explore the wider aspects of organizational agility. This 7. Conclusion
expanded investigation could uncover additional dimensions and po
tential intersections that could enhance our understanding of how AI- Our research investigates how AI-powered capabilities can add value
powered capabilities contribute to firms’ responsiveness and adapt to organizations through their capacity to directly affect how firms
ability. Further, as Kalaignanam et al. (2021) observed, sensemaking, identify and respond to market changes. We offer a concept of ARMC
speed, and iteration define agility. Our concept of ARMC does not derived from organizational agility and dynamic capabilities perspec
explicitly contain iteration; however, it aligns with the crucial elements tives. We also investigate the moderating role of environmental condi
of sensemaking and speed, which are integral components of organiza tions in influencing the relationship between AI-powered capabilities
tional agility. We call for future research to conduct a longitudinal study, and ARMC. Our findings indicate that the effects of AI-enabled auto
explicitly observing all three aspects of agility in the study. mation, AI-enabled analytics, and AI-enabled relational capability on
Second, while many researchers have defined agility as encompass ARMC are contingent on two environmental conditions—environmental
ing speed and flexibility (e.g., Kalaignanam et al., 2021), Prange (2021) hostility and environmental dynamism. Our research provides a theory-
suggests that agility is not reliant solely on speed as a defining charac based understanding of AI’s impact on organizations. In addition, we
teristic. Specifically, according to Prange, agility requires both the endeavor to establish a foundational framework for future theory-driven
resilience derived from a reliable core and the dynamism anchored in studies to gain a comprehensive understanding of how AI benefits or
ongoing change. Our conceptualization of ARMC does not explicitly ganizations. This understanding can contribute to improved organiza
elaborate on the concept of agility as slowness. Future research should tional performance and foster additional competitive actions.
explore different aspects of agility to lead to a better understanding of
how agility, including ARMC, can contribute to organizational success in CRediT authorship contribution statement
varying environments.
Third, we conceptualize each AI-powered capability as a first-order Yulia Sullivan: Writing – review & editing, Writing – original draft,
construct by following the existing concept of multiple AIs (e.g., Methodology, Formal analysis, Data curation, Conceptualization.
Huang et al., 2019; Huang & Rust, 2021a). However, the relationships Samuel Fosso Wamba: Writing – original draft, Resources, Project
among AI-powered capabilities may be more complex than we have administration, Methodology, Investigation, Funding acquisition,
hypothesized. For instance, interaction effects could occur among AI Conceptualization.
capabilities. A larger sample size would be required to test such complex
relationships. To examine these complexities further, exploratory Declaration of competing interest
research could achieve a much stronger theoretical foundation. Under
standing the dependencies among a wider capability system in an or The authors declare that they have no known competing financial
ganization could be an interesting and important agenda for future interests or personal relationships that could have appeared to influence
research in this area, especially under different environmental the work reported in this paper.
conditions.
Fourth, although we conducted a two-stage survey to address the Data availability
threats of common method bias, we still used the same respondent for
both our independent and dependent variables. Although common Data will be made available on request.
Appendix A. . Measurement items
AI-Enabled Automatic Capability (New scale, measured on a seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree).
1. In our firm, AI tools and applications are used to automate time-consuming activities, including claim processing, basic customer-service inter
action, and inventory tracking.
2. In our firm, AI tools and applications are used to save staff time by allowing them to focus on higher-value activities.
3. In our firm, AI tools and applications are used to automate routine as well as back-office tasks.
AI-Enabled Analytics Capability (New scale, measured on a seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree).
In our organization, AI tools and applications allow us to….
1. …predict more accurately potential equipment or service failures.

2. …respond more quickly to changing stock or inventory levels.
3. …test nearly all potential scenarios before making a decision.
4. …rapidly analyze millions of combinations to help employees test the efficiency of different decisions under different conditions.
13
5. …for our executive teams understand likely ranges of outcomes resulting from acquisitions of suppliers.
6. …accurately price our products or services (dropped)
7. …predict more accurately the next-best product or service a customer is likely to buy.
8. …forecast our customer traffic (e.g., in our stores or on our websites).
9. …for our middle managers simulate product or service delivery to identify high-value products or services.
AI-Enabled Relational Capability (New scale, adapted from Lee et al., 2020, measured on a seven-point Likert scale ranging from 1 = strongly
disagree to 7 = strongly agree).
1. Our AI-based system offers various decision-making tools that enable us to manage our relationship with our customers.
2. Our AI-based system offers various tools, such as chatbots, that enable us to support our interactions with our customers.
3. Our AI-based system offers various tools that enable us to examine trends in the data for managing our interaction with our customers.
4. We use virtual agents (e.g., chatbots and machine learning) to respond to our customers’ queries.
Adaptive Response to Market Changes (ARMC) (Source: Tallon & Pinsonneault, 2011).
How easily and quickly can your firm perform the following actions? (measured on a seven-point Likert scale ranging from 1 = strongly disagree to
7 = strongly agree).
Customer responsiveness
1. Respond to changes in aggregate consumer demand.

2. Customize a product or service to suit an individual customer.
3. React to new product or service launches by competitors.
4. Introduce new pricing schedules in response to changes in competitors’ or markets’ prices.
Operation adjustment
1. Expand into new regional or international markets.

2. Change (i.e., expand or reduce) the variety of products/services available for sale.
3. Adopt new technologies to produce better, faster, and cheaper products and services.
4. Switch suppliers to avail of lower costs, better quality, or improved delivery times (dropped)
Environmental Hostility (Source: Chen et al., 2014, measured on a seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly
agree).
1. The survival of our firm is currently threatened by the scarce supply of labor.
2. The survival of our firm is currently threatened by the scarce supply of materials.
3. The survival of our firm is currently threatened by tough price competition.
4. The survival of our firm is currently threatened by tough competition in product/service quality.
5. The survival of our firm is currently threatened by tough competition in product/service differentiation.
Environmental Dynamism (Source: Lee et al., 2015, measured on a seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly
agree).
1. Technological changes in our industry are rapid.

2. Customers’ product/service preferences change rapidly.
3. The actions of competitors in our major markets are changing rapidly (dropped)
Innovation (Source: Prajogo & Ahmed, 2006).

Indicate your firm’s ability to innovate during the COVID-19 crisis relative to all other competitors in the same industry (1 = far below average to 7
= far above average) (note: the COVID-19 statement was used in the questionnaire because data were collected during the COVID pandemic).
Product innovation
1. Level of newness (novelty)

2. Use of the latest technology
3. Speed of product development (dropped)
4. Number of new products/services
Process innovation
1. Technology competitiveness
2. Speed of adopting the latest technology
3. Novelty of the technology use
14
Firm Performance (Source: Queiroz et al., 2018, measured on a seven-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree).
To what extent do the following statements reflect your firm’s current situation in the last 12 months?
1. We are more profitable than our competitors.

2. Our sales growth exceeds that of our competitors.
3. Our revenue growth exceeds that of our competitors.
4. Our market share growth exceeds that of our competitors.
5. Overall, our performance is better than our competitors.
Appendix B. . WarpPLS loadings and Cross-Loadings
AI_RL AI_ANL AI_AUTO ARMC E_HOST E_DYN PROC_INN PROD_INN FP IND C_AGE C_SIZE LOC
AI_RL1 0.87 − 0.02 − 0.07 0.02 0.17 0.11 − 0.01 0.01 − 0.03 − 0.04 0.11 0.01 − 0.07
AI_RL2 0.89 − 0.03 0.13 − 0.08 − 0.04 0.09 0.02 − 0.10 0.09 0.06 − 0.05 0.03 0.03
AI_RL3 0.91 0.27 − 0.23 0.09 − 0.10 − 0.05 0.05 − 0.08 − 0.02 − 0.08 0.03 − 0.13 0.04
AI_RL4 0.81 − 0.25 0.20 − 0.03 − 0.04 − 0.16 − 0.05 0.19 − 0.04 0.08 − 0.10 0.10 − 0.01
AI_ANL1 − 0.24 0.88 0.14 − 0.18 0.01 0.38 0.04 − 0.05 0.04 0.04 − 0.05 0.17 − 0.07
AI_ANL2 0.16 0.86 0.20 − 0.21 0.04 0.09 − 0.03 0.07 0.02 0.08 0.04 0.01 − 0.03
AI_ANL3 − 0.22 0.87 0.19 − 0.03 − 0.03 − 0.25 − 0.16 0.18 − 0.09 − 0.05 0.02 − 0.02 0.05
AI_ANL4 0.06 0.81 0.38 0.00 − 0.06 − 0.21 − 0.27 0.26 − 0.08 0.01 − 0.09 0.03 0.05
AI_ANL5 0.21 0.84 − 0.33 0.28 − 0.01 0.03 0.00 0.07 − 0.08 − 0.04 0.18 − 0.06 − 0.14
AI_ANL7 0.02 0.90 − 0.09 − 0.03 0.06 0.08 0.11 − 0.16 0.11 − 0.04 − 0.01 − 0.09 − 0.01
AI_ANL8 − 0.11 0.84 − 0.26 0.17 − 0.08 − 0.16 0.05 − 0.09 0.01 − 0.07 − 0.11 − 0.12 0.10
AI_ANL9 0.14 0.89 − 0.22 0.01 0.06 0.01 0.24 − 0.25 0.06 0.07 0.01 0.07 0.04
AI_AUTO1 0.15 − 0.41 0.85 − 0.05 − 0.03 0.01 0.07 − 0.08 0.05 0.16 − 0.09 0.08 0.10
AI_AUTO2 0.07 − 0.14 0.92 0.07 − 0.03 − 0.04 − 0.12 0.10 − 0.01 0.00 0.06 0.01 − 0.06
AI_AUTO3 − 0.25 0.63 0.76 − 0.03 0.07 0.05 0.07 − 0.03 − 0.05 − 0.17 0.02 − 0.10 − 0.04
ARMC_CR − 0.08 0.23 − 0.21 0.92 − 0.03 0.05 − 0.12 0.12 0.00 − 0.06 0.02 − 0.06 − 0.07
ARMC_OA 0.08 − 0.23 0.21 0.92 0.03 − 0.05 0.12 − 0.12 0.00 0.06 − 0.02 0.06 0.07
(continued on next page)
Fig. C1. 3D Plots for moderating effects.
15
(continued )
AI_RL AI_ANL AI_AUTO ARMC E_HOST E_DYN PROC_INN PROD_INN FP IND C_AGE C_SIZE LOC
E_HOST1 − 0.25 0.31 − 0.11 0.02 0.88 0.03 − 0.10 0.10 − 0.04 − 0.06 − 0.11 0.08 0.09
E_HOST2 − 0.06 0.24 − 0.14 − 0.14 0.88 0.13 − 0.06 0.03 0.11 − 0.09 − 0.06 0.08 0.05
E_HOST3 0.17 − 0.39 0.21 − 0.10 0.90 0.13 0.05 − 0.18 0.08 0.05 0.07 − 0.09 − 0.10
E_HOST4 0.03 0.05 − 0.10 0.15 0.92 − 0.18 − 0.05 0.19 − 0.16 − 0.02 0.01 − 0.03 0.03
E_HOST5 0.10 − 0.20 0.13 0.07 0.90 − 0.10 0.15 − 0.14 0.01 0.11 0.10 − 0.04 − 0.07
E_DYN1 0.00 − 0.34 0.09 − 0.17 − 0.05 0.88 0.03 0.05 − 0.06 − 0.01 − 0.02 0.07 − 0.05
E_DYN2 0.00 0.34 − 0.09 0.17 0.05 0.88 − 0.03 − 0.05 0.06 0.01 0.02 − 0.07 0.05
PROC_INN1 0.14 0.03 − 0.19 0.07 − 0.01 − 0.02 0.93 0.01 − 0.05 0.01 0.02 0.01 0.03
PROC_INN2 − 0.10 − 0.10 0.12 − 0.04 0.00 0.05 0.90 − 0.05 − 0.04 − 0.01 0.01 − 0.02 − 0.04
PROC_INN3 − 0.04 0.07 0.08 − 0.04 0.01 − 0.04 0.90 0.04 0.09 − 0.01 − 0.03 0.01 0.01
PROD_INN1 0.14 0.08 − 0.15 0.05 − 0.01 − 0.08 − 0.23 0.87 0.23 − 0.09 0.03 0.10 − 0.09
PROD_INN2 0.10 − 0.15 0.00 0.04 − 0.07 0.03 0.36 0.86 − 0.30 0.10 0.02 0.07 − 0.08
PROD_INN4 − 0.24 0.07 0.15 − 0.08 0.08 0.05 − 0.14 0.88 0.07 − 0.01 − 0.05 − 0.17 0.18
FP1 − 0.13 − 0.06 0.18 0.01 − 0.01 − 0.08 0.16 − 0.26 0.88 − 0.10 − 0.03 − 0.02 0.02
FP2 0.00 0.03 − 0.06 0.08 0.03 − 0.07 0.22 − 0.22 0.93 0.02 0.04 − 0.04 − 0.04
FP3 − 0.11 − 0.07 0.09 − 0.09 0.15 0.11 − 0.21 0.07 0.90 − 0.07 0.05 − 0.09 0.00
FP4 0.19 0.09 − 0.17 − 0.01 − 0.11 0.02 − 0.19 0.29 0.85 0.14 − 0.08 0.15 0.04
FP5 0.06 0.01 − 0.03 0.01 − 0.07 0.01 0.01 0.14 0.90 0.02 0.02 0.00 − 0.03
Appendix C. . 3D Plots for the moderating effects of environmental conditions
Fig. C1.
References media marketing with potential users’ expectations. Technological Forecasting and
Social Change, 151, Article 119794.
Chakravarty, A., Grewal, R., & Sambamurthy, V. (2013). Information technology
Albrecht, T., Rausch, T. M., & Derra, N. D. (2021). Call me maybe: Methods and practical
competencies, organizational agility, and firm performance: Enabling and
implementation of artificial intelligence in call center arrivals’ forecasting. Journal of
facilitating roles. Information Systems Research, 24(4), 976–997.
Business Research, 123, 267–278.
Chatterjee, L., Feng, C., Nakata, C., & Sivakumar, K. (2023). The environmental
Allal-Chérif, O., Simón-Moya, V., & Ballester, A. C. C. (2021). Intelligent purchasing:
turbulence concept in marketing: A look back and a look ahead. Journal of Business
How artificial intelligence can redefine the purchasing function. Journal of Business
Research, 161, Article 113775.
Research, 124, 69–76.
Cheatham, B., Cosmas, A., Mehta, N., & Shah, D. (2020). How to build AI with (and for)
Anand, G., & Ward, P. T. (2004). Fit, flexibility and performance in manufacturing:
everyone in your organization. Retrieved From: McKinsey & Company. https://www.
Coping with dynamic environments. Production and Operations Management, 13(4),
mckinsey.com/business-functions/mckinsey-analytics/our-insights/how-to-build-ai-
369–385.
with-and-for-everyone-in-your-organization.
Armstrong, J. S., & Overton, T. S. (1977). Estimating nonresponse bias in mail surveys.
Chen, D. Q., Preston, D. S., & Swink, M. (2021). How big data analytics affects supply
Journal of Marketing Research, 14(3), 396–402.
chain decision-making: An empirical analysis. Journal of the Association for
Asatiani, A., Malo, P., Nagbøl, P. R., Penttinen, E., Rinta-Kahila, T., & Salovaara, A.
Information Systems, 22(5), 1224–1244.
(2020). Challenges of explaining the behavior of black-box AI systems. MIS Quarterly
Chen, Y., Wang, Y., Nevo, S., Benitez, J., & Kou, G. (2017). Improving strategic flexibility
Executive, 19(4), 259–278.
with information technologies: Insights for firm performance in an emerging
Barua, A., Konana, P., Whinston, A. B., & Yin, F. (2004). An empirical investigation of
economy. Journal of Information Technology, 32(1), 10–25.
net-enabled business value. MIS Quarterly, 28(4), 585–620.
Chen, Y., Wang, Y., Nevo, S., Jin, J., Wang, L., & Chow, W. S. (2014). IT capability and
Battleson, D. A., West, B. C., Kim, J., Ramesh, B., & Robinson, P. S. (2016). Achieving
organizational performance: The roles of business process agility and environmental
dynamic capabilities with cloud computing: An empirical investigation. European
factors. European Journal of Information Systems, 23(3), 326–342.
Journal of Information Systems, 25(3), 209–230.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed). Hillsdale,
Benbya, H., Davenport, T. H., & Pachidi, S. (2020). Artificial intelligence in
NJ: Erlbaum.
organizations: Current state and future opportunities. MIS Quarterly Executive, 19(4),
Côrte-Real, N., Ramalho, N., Oliveira, T., & Morgado, E. M. (2017). Digital business
ix–xxi.
transformation and strategy: What do we know so far? A systematic literature review
Benbya, H., Pachidi, S., & Jarvenpaa, S. (2021). Special issue editorial: Artificial
and research agenda. International Journal of Information Management, 37(6),
intelligence in organizations: Implications for information systems research. Journal
557–570.
of the Association for Information Systems, 22(2), 281–303.
Damanpour, F. (1991). Organizational innovation: A meta-analysis of effects of
Benner, M. J., & Tushman, M. L. (2015). Reflections on the 2013 Decade
determinants and moderators. Academy of Management Journal, 34, 555–590.
Award—“Exploitation, exploration, and process management: The productivity
Damanpour, F., & Gopalakrishnan, S. (2001). The dynamics of the adoption of product
dilemma revisited” ten years later. Academy of Management Review, 40(4), 497–514.
and process innovations in organizations. Journal of Management Studies, 38(1),
Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Managing artificial intelligence.
45–65.
MIS Quarterly, 45(3), 1433–1450.
Davenport, T. H. (2018). From analytics to artificial intelligence. Journal of Business
Bhattacherjee, A., & Hikmet, N. (2007). Physicians’ resistance toward healthcare
Analytics, 1(2), 73–80.
information technology: A theoretical model and empirical test. European Journal of
Davenport, T. H., Guha, A., Grewal, D., & Bressgott, T. (2020). How artificial intelligence
Information Systems, 16(6), 725–737.
will change the future of marketing. Journal of the Academy of Marketing Science, 48
Bhoovaraghavan, S., Vasudevan, A., & Chandran, R. (1996). Resolving the process vs.
(1), 24–42.
product innovation dilemma: A consumer choice theoretic approach. Management
Davenport, T. H., & Mahidhar, V. (2018). What’s your cognitive strategy. MIT Sloan
Science, 42(2), 232–246.
Management Review, 59(4), 19–23.
Bourgeois, & Eisenhardt, K. M. (1988). Strategic decision processes in high velocity
Davenport, T. H., & Ronanki, R. (2018). Artificial intelligence for the real world. Harvard
environments: Four cases in the microcomputer industry. Management Science, 34(7),
Business Review, 96(1), 108–116.
816–835.
Deloitte.. (2020). Automation with intelligence: Pursuing organization-wide reimagination.
Bradley, R. V., Pratt, R. M., Byrd, T. A., Outlay, C. N., & Wynn, D. E., Jr (2012).
Retrieved From: Deloitte Insights. https://www2.deloitte.com/content/dam/
Enterprise architecture, IT effectiveness and the mediating role of IT alignment in US
insights/us/articles/73699-global-intelligent-automation-survey/DI_Automation-
hospitals. Information Systems Journal, 22(2), 97–127.
with-intelligence.pdf.
Brock, J. K. U., & Von Wangenheim, F. (2019). Demystifying AI: What digital
Dess, G. G., & Beard, D. W. (1984). Dimensions of organizational task environments.
transformation leaders can teach you about realistic artificial intelligence. California
Administrative Science Quarterly, 29(1), 52–73.
Management Review, 61(4), 110–134.
Doherty, N. F., & Terry, M. (2009). The role of IS capabilities in delivering sustainable
Butner, K., Ivory, T., Albertoni, M., & Sotheran, K. (2020). Automation and the future of
improvements to competitive positioning. Journal of Strategic Information Systems, 18
work. From: IBM Research Insights. https://www.ibm.com/downloads/cas
(2), 100–116.
/O17AWNV6.
Capatina, A., Kachour, M., Lichy, J., Micu, A., Micu, A. E., & Codignola, F. (2020).
Matching the future capabilities of an artificial intelligence-based software for social
16
Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Artificial intelligence for decision Kohli, R., & Grover, V. (2008). Business value of IT: An essay on expanding research
making in the era of Big Data–evolution, challenges, and research agenda. directions to keep up with the times. Journal of the Association for Information
International Journal of Information Management, 48, 63–71. Systems, 9(1), 23-39.
Eisenhardt, K. M., & Martin, J. A. (2000). Dynamic capabilities: What are they? Strategic Kulkarni, U., Robles-Flores, J. A., & Popovič, A. (2017). Business intelligence capability:
Management Journal, 21(10–11), 1105–1121. The effect of top management and the mediating roles of user participation and
Elazhary, M., Popovič, A., de Souza, H., Bermejo, P., & Oliveira, T. (2023). How analytical decision making orientation. Journal of the Association for Information
information technology governance influences organizational agility: The role of Systems, 18(7), 516–541.
market turbulence. Information Systems Management, 40(2), 148–168. Kyläheiko, K., & Maijanen, P. (2020). Industry 4.0. transformation challenge in light of
Fernández-Rovira, C., Valdés, J.Á., Molleví, G., & Nicolas-Sans, R. (2021). The digital dynamic capabilities. Technical, Economic and Societal Effects of Manufacturing, 4,
transformation of business. Towards the datafication of the relationship with 169–189.
customers. Technological Forecasting and Social Change, 162, Article 120339. Lee, N. C. A., Wang, E. T., & Grover, V. (2020). IOS drivers of manufacturer-supplier
Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable flexibility and manufacturer agility. Journal of Strategic Information Systems, 29(1),
variables and measurement error: Algebra and statistics. Journal of Marketing Article 101594.
Research, 18(3), 382–388. Lee, O. K., Sambamurthy, V., Lim, K. H., & Wei, K. K. (2015). How does IT ambidexterity
Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing moderator and mediator effects impact organizational agility? Information Systems Research, 26(2), 398–417.
in counseling psychology research. Journal of Counseling Psychology, 51(1), 115. Lin, C., & Kunnathur, A. (2019). Strategic orientations, developmental culture, and big
Georgsdottir, A. S., & Getz, I. (2004). How flexibility facilitates innovation and ways to data capability. Journal of Business Research, 105, 49–60.
manage it in organizations. Creativity and Innovation Management, 13(3), 166–175. Lindell, M. K., & Whitney, D. J. (2001). Accounting for common method variance in
Gregory, R. W., Henfridsson, O., Kaganer, E., & Kyriakou, H. (2021). The role of artificial cross-sectional research designs. Journal of Applied Psychology, 86(1), 114.
intelligence and data network effects for creating user value. Academy of Management Lu, Y., & Ramamurthy, K. (2011). Understanding the link between information
Review, 46(3), 534–551. technology capability and organizational agility: An empirical examination. MIS
Grover, V., Kohli, R., & Guo, C. (2018). Business value of IT in the digital era: A study of Quarterly, 35(4), 931–954.
IT-enabled digital transformation. Journal of Management Information Systems, 35(3), Luo, X., Tong, S., Fang, Z., & Qu, Z. (2019). Frontiers: Machines vs. humans: The impact
984–1015. of artificial intelligence chatbot disclosure on customer purchases. Marketing Science,
Gupta, S., Modgil, S., Bhattacharyya, S., & Bose, I. (2022). Artificial intelligence for 38(6), 937–947.
decision support systems in the field of operations research: Review and future scope Makarius, E. E., Mukherjee, D., Fox, J. D., & Fox, A. K. (2020). Rising with the machines:
of research. Annals of Operations Research, 308, 1–60. A sociotechnical framework for bringing artificial intelligence into the organization.
Haefner, N., Wincent, J., Parida, V., & Gassmann, O. (2021). Artificial intelligence and Journal of Business Research, 120, 262–273.
innovation management: A review, framework, and research agenda. Technological Malhotra, N. K., Kim, S. S., & Patil, A. (2006). Common method variance in IS research: A
Forecasting and Social Change, 162, Article 120392. comparison of alternative approaches and a reanalysis of past research. Management
Helfat, C. E. (2022). Strategic organization, dynamic capabilities, and the external Science, 52(12), 1865–1883.
environment. Strategic Organization, 20(4), 734–742. Manis, K. T., & Madhavaram, S. (2023). AI-Enabled marketing capabilities and the
Helfat, C. E., Finkelstein, S., Mitchell, W., Peteraf, M. A., Singh, H., Teece, D. J., & hierarchy of capabilities: Conceptualization, proposition development, and research
Winter, S. G. (2007). Dynamic Capabilities: Understanding Strategic Change in avenues. Journal of Business Research, 157, Article 113485.
Organizations. Carlton: Blackwell. Mathieson, K., Peacock, E., & Chin, W. W. (2001). Extending the technology acceptance
Helfat, C. E., & Peteraf, M. A. (2015). Managerial cognitive capabilities and the model: The influence of perceived user resources. ACM SIGMIS Database: the DATABASE
microfoundations of dynamic capabilities. Strategic Management Journal, 36(6), for Advances in Information Systems, 32(3), 86–112.
831–850. Mellahi, K. (2001). Differences and similarities in future managerial values: A five
Hisrich, R. D. (1990). Entrepreneurship/Intrapreneurship. American Psychologist, 45, cultures comparative study. Cross Cultural Management: An International Journal, 8
209–222. (1), 45–58.
Hollebeek, L. D., Sprott, D. E., & Brady, M. K. (2021). Rise of the machines? Customer Melville, N., Kraemer, K., & Gurbaxani, V. (2004). Information technology and
engagement in automated service interactions. Journal of Service Research, 24(1), organizational performance: An integrative model of IT business value. MIS
3–8. Quarterly, 28(2), 283–322.
Holmqvist, M., & Pessi, K. (2006). Agility through scenario development and continuous Mitchell, J. R., Shepherd, D. A., & Sharfman, M. P. (2011). Erratic strategic decisions:
implementation: A global aftermarket logistics case. European Journal of Information When and why managers are inconsistent in strategic decision making. Strategic
Systems, 15(2), 146–158. Management Journal, 32(7), 683–704.
Hoonsopon, D., & Puriwat, W. (2019). Organizational agility: Key to the success of new Moore, G. C., & Benbasat, I. (1991). Development of an instrument to measure the
product development. IEEE Transactions on Engineering Management, 68(6), perceptions of adopting an information technology innovation. Information Systems
1722–1733. Research, 2(3), 192–222.
Huang, M. H., & Rust, R. T. (2021a). Engaged to a robot? The role of AI in service. Morgan, N. A., Vorhies, D. W., & Mason, C. H. (2009). Market orientation, marketing
Journal of Service Research, 24(1), 30–41. capabilities, and firm performance. Strategic Management Journal, 30(8), 909–920.
Huang, M. H., & Rust, R. T. (2021b). A strategic framework for artificial intelligence in Mumford, M. D., & Gustafson, S. B. (1988). Creativity syndrome: Integration, application
marketing. Journal of the Academy of Marketing Science, 49(1), 30–50. and innovation. Psychological Bulletin, 103, 27–43.
Huang, M. H., & Rust, R. T. (2022). A framework for collaborative artificial intelligence Nambisan, S., Lyytinen, K., Majchrzak, A., & Song, M. (2017). Digital innovation
in marketing. Journal of Retailing, 98(2), 209–223. management: Reinventing innovation management research in a digital world. MIS
Huang, M. H., Rust, R., & Maksimovic, V. (2019). The feeling economy: Managing in the Quarterly, 41(1), 223–238.
next generation of artificial intelligence (AI). California Management Review, 61(4), Narayandas, D., & Sengupta, A. (2023). Using AI to adjust your marketing and sales in a
43–65. volatile world. Harvard Business Review. April 12. Retrieved from https://hbr.org/
Hulland, J., Baumgartner, H., & Smith, K. M. (2018). Marketing survey research best 2023/04/using-ai-to-adjust-your-marketing-and-sales-in-a-volatile-world.
practices: Evidence and recommendations from a review of JAMS articles. Journal of Nazir, S., & Pinsonneault, A. (2012). IT and firm agility: An electronic integration
the Academy of Marketing Science, 46, 92–108. perspective. Journal of the Association for Information Systems, 13(3), 150–171.
Iansati, M., & Lakhari, K. R. (2020). Competing in the age of AI. Harvard Business Review. Ngai, E. W., Chau, D. C., & Chan, T. L. A. (2011). Information technology, operational,
January-February. and management competencies for supply chain agility: Findings from case studies.
Im, S., & Workman, J. P., Jr. (2004). Market orientation, creativity, and new product Journal of Strategic Information Systems, 20(3), 232–249.
performance in high-technology firms. Journal of Marketing, 68(2), 114–132. Oke, A., Burke, G., & Myers, A. (2007). Innovation types and performance in growing UK
Jaworski, B. J., & Kohli, A. K. (1993). Market orientation: Antecedents and consequences. SMEs. International Journal of Operations & Production Management, 27(7), 735–753.
Journal of Marketing, 57(3), 53–70. Organization for Economic Co-operation and Development (OECD). (2019). ICT access
Jussupow, E., Spohrer, K., Heinzl, A., & Gawlitza, J. (2021). Augmenting medical and usage by households and individuals. From: https://stats.oecd.org/Index.aspx?
diagnosis decisions? An investigation into physicians’ decision-making process with DataSetCode=ICT_HH2.
artificial intelligence. Information Systems Research, 32(3), 713–735. Özsomer, A., Calantone, R. J., & Di Bonetto, A. (1997). What makes firms more
Kalaignanam, K., Tuli, K. R., Kushwaha, T., Lee, L., & Gal, D. (2021). Marketing agility: innovative? A look at organizational and environmental factors. Journal of Business &
The concept, antecedents, and a research agenda. Journal of Marketing, 85(1), 35–58. Industrial Marketing, 12(6), 400–416.
Kaplan, A., & Haenlein, M. (2019). Siri, Siri, in my hand: Who’s the fairest in the land? Paschen, U., Pitt, C., & Kietzmann, J. (2020). Artificial intelligence: Building blocks and
On the interpretations, illustrations, and implications of artificial intelligence. an innovation typology. Business Horizons, 63(2), 147–155.
Business Horizons, 62(1), 15–25. Pavlou, P. A., & El Sawy, O. A. (2006). From IT leveraging competence to competitive
Kathuria, A., Mann, A., Khuntia, J., Saldanha, T. J., & Kauffman, R. J. (2018). A strategic advantage in turbulent environments: The case of new product development.
value appropriation path for cloud computing. Journal of Management Information Information Systems Research, 17(3), 198–227.
Systems, 35(3), 740–775. Pavlou, P. A., & El Sawy, O. A. (2011). Understanding the elusive black box of dynamic
Kiron, D., & Schrage, M. (2019). Strategy for and with AI. MIT Sloan Management Review, capabilities. Decision Sciences, 42(1), 239–273.
60(4), 30–35. Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Common method
Kock, N. (2022). WarpPLS user manual: Version 8.0. ScriptWarp Systems: Laredo, TX, biases in behavioral research: A critical review of the literature and recommended
USA. remedies. Journal of Applied Psychology, 88(5), 879–903.
Kohli, A. K., & Jaworski, B. J. (1990). Market orientation: The construct, research Polites, G. L., & Karahanna, E. (2012). Shackled to the status quo: The inhibiting effects
propositions, and managerial implications. Journal of Marketing, 54(2), 1–18. of incumbent system habit, switching costs, and inertia on new system acceptance.
MIS Quarterly, 36(1), 21–42.
17
Prajogo, D. I., & Ahmed, P. K. (2006). Relationships between innovation stimulus, Tussyadiah, I. (2020). A review of research into automation in tourism: Launching the
innovation capacity, and innovation performance. R&D Management, 36(5), annals of tourism research curated collection on artificial intelligence and robotics in
499–515. tourism. Annals of Tourism Research, 81, Article 102883.
Prange, C. (2021). Agility as the discovery of slowness. California Management Review, 63 Wang, C. L., & Ahmed, P. K. (2004). The development and validation of the
(4), 27–51. organizational innovativeness construct using confirmatory factor analysis. European
Puriwat, W., & Hoonsopon, D. (2021). Cultivating product innovation performance Journal of Innovation Management, 7(4), 303–313.
through creativity: The impact of organizational agility and flexibility under Wang, W. T., Chih, W. H., & Jiang, J. J. (2012). Analyzing the determinants of business
technological turbulence. Journal of Manufacturing Technology Management, 33(4), value creation in e-business settings. Information & Management, 49(8), 385–397.
741–762. Whittlestone, J., Nyrup, R., Alexandrova, A., Dihal, K., & Cave, S. (2019). Ethical and
Queiroz, M., Tallon, P. P., Sharma, R., & Coltman, T. (2018). The role of IT application Societal Implications of Algorithms, Data, and Artificial Intelligence: A Roadmap for
orchestration capability in improving agility and performance. Journal of Strategic Research. London: Nuffield Foundation.
Information Systems, 27(1), 4–21. Wilden, R., & Gudergan, S. P. (2015). The impact of dynamic capabilities on operational
Raju, P., & Koch, R. (2019). Can RPA improve agility? Strategic Finance (March), 68–69. marketing and technological capabilities: Investigating the role of environmental
Ravichandran, T. (2018). Exploring the relationships between IT competence, innovation turbulence. Journal of the Academy of Marketing Science, 43, 181–199.
capacity and organizational agility. Journal of Strategic Information Systems, 27(1), Zahra, S. A., & Garvis, D. M. (2000). International corporate entrepreneurship and firm
22–42. performance: The moderating effect of international environmental hostility. Journal
Ravichandran, T., Lertwongsatien, C., & Lertwongsatien, C. (2005). Effect of information of Business Venturing, 15(5–6), 469–492.
systems resources and capabilities on firm performance: A resource-based Zhou, K. Z., & Wu, F. (2010). Technological capability, strategic flexibility, and product
perspective. Journal of Management Information Systems, 21(4), 237–276. innovation. Strategic Management Journal, 31(5), 547–561.
Rosenblat, A. (2018). Uberland: How Algorithms are Rewriting the Rules of Work. Univ of
California Press.
Yulia Sullivan, Ph.D. is an Associate Professor of Information Systems and Business
Sambamurthy, V., Bharadwaj, A., & Grover, V. (2003). Shaping agility through digital
Analytics in Hankamer Business School at Baylor University. She earned her Ph.D. degree
options: Reconceptualizing the role of information technology in contemporary
in information systems from the University of North Texas. Her research interests include
firms. MIS Quarterly, 27(2), 237–263.
cognitive information systems, the value of IT for organizations, artificial intelligence &
Shrestha, Y. R., Ben-Menahem, S. M., & Von Krogh, G. (2019). Organizational decision-
ethics, human-computer interaction, and research methods. Her work has been published
making structures in the age of artificial intelligence. California Management Review,
in various journals, including MIS Quarterly, Journal of the Association for Information
61(4), 66–83.
Systems, Journal of Business Ethics, International Journal of Information Management, Journal
Slevin, D. P., & Covin, J. G. (1997). Strategy formation patterns, performance, and the
of Medical Internet Research, and others. She has also presented her research at many in
significance of context. Journal of Management, 23(2), 189–209.
ternational conferences, including the International Conference on Information Systems
Straub, D. W. (1989). Validating instruments in MIS research. MIS Quarterly, 13(2),
and the Hawaii International Conference on System Sciences. She is also a co-author of
147–169.
Conducting Mixed-Methods Research: From Classical Social Sciences to the Age of Big Data
Tallon, P. P., & Pinsonneault, A. (2011). Competing perspectives on the link between
and Analytics.
strategic information technology alignment and organizational agility: Insights from
a mediation model. MIS Quarterly, 35(2), 463–486.
Tan, T. F., Netessine, S., & Hitt, L. (2017). Is Tom Cruise threatened? An empirical study Samuel Fosso Wamba, Ph. D. is a Professor at TBS Education, France, and a Distinguished
of the impact of product variety on demand concentration. Information Systems Visiting Professor at The University of Johannesburg, South Africa. He earned his Ph.D. in
Research, 28(3), 643–660. industrial engineering at the Polytechnic School of Montreal, Canada. He earned his Ph.D.
Tanriverdi, H., & Venkatraman, N. (2005). Knowledge relatedness and the performance in industrial engineering at the Polytechnic School of Montreal, Canada. He also holds two
of multibusiness firms. Strategic Management Journal, 26(2), 97–119. master’s degrees: one MSc in mathematics from the University of Sherbrooke in Canada
Teece, D. J. (2007). Explicating dynamic capabilities: The nature and microfoundations and another MSc in electronic commerce with a major in computer science from HEC
of (sustainable) enterprise performance. Strategic Management Journal, 28(13), Montreal, Canada. His current research focuses on the business value of I.T., inter-
1319–1350. organizational systems adoption, use and impacts, supply chain management, electronic
Teece, D. J. (2014). The foundations of enterprise performance: Dynamic and ordinary commerce, blockchain, artificial intelligence for business, social media, business analytics,
capabilities in an (economic) theory of firms. Academy of Management Perspectives, 28 big data, and open data. He is a CompTIA RFID+ Certified Professional and the Academic
(4), 328–352. Co-Founder of RFID Academia. Apart from teaching and conducting research, He leads the
Teece, D., Peteraf, M., & Leih, S. (2016). Dynamic capabilities and organizational agility: newly created Center of Excellence in Artificial Intelligence & Business Analytics at TBS
Risk, uncertainty, and strategy in the innovation economy. California Management Education. He has been recently recognized as the most influential scholar in big data
Review, 58(4), 13–35. analytics and enterprises based on the number of published articles and citations and
Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic among the 2% of the most influential scholars in the world based on the Mendeley data
management. Strategic Management Journal, 18(7), 509–533. base that includes 100,000 top scientists. He ranks in ClarivateTMs 1% most cited scholars
Trantopoulos, K., von Krogh, G., Wallin, M. W., & Woerter, M. (2017). External in the world for 2020, from the “Highly Cited ResearchersTM” list that identifies global
knowledge and information technology. MIS Quarterly, 41(1), 287–300. research scientists who show exceptional influence-reflected in the publication of multiple
Tushman, M. L., & Nadler, D. (1986). Organizing for innovation. California Management papers frequently cited by their peers. He ranks in CDO Magazine’s Leading Academic
Review, 28(3), 74–92. Data Leaders 2021.
18

AI and Firm Performance

Uploaded by

Copyright:

Available Formats

AI and Firm Performance

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AI and Firm Performance

Uploaded by

Copyright:

Available Formats

Journal of Business Research 174 (2024) 114500

Contents lists available at ScienceDirect

Journal of Business Research

Artificial intelligence and adaptive response to market changes: A strategy

1. Introduction introduces a new set of challenges to this longstanding issue. Conse­

access and functional-related competencies (Battleson et al., 2016; Table 1

Fig. 1. Conceptual model.

High Env. Low ARMC High ARMC Low ARMC

1 AI-enabled automation 0.85

Fig. 2. Structural model.

Model 1a Model 1b Model 2a Model 2b Model 3a Model 3b Model 4a Model 4b

Adj. R2 0.53 0.68 0.00 0.13 0.03 0.09 0.04 0.13

Table 7 Table 7 (continued )

Appendix A. . Measurement items

1. …predict more accurately potential equipment or service failures.

1. Respond to changes in aggregate consumer demand.

1. Expand into new regional or international markets.

1. Technological changes in our industry are rapid.

Innovation (Source: Prajogo & Ahmed, 2006).

1. Level of newness (novelty)

1. We are more profitable than our competitors.

Appendix B. . WarpPLS loadings and Cross-Loadings

Fig. C1. 3D Plots for moderating effects.

Appendix C. . 3D Plots for the moderating effects of environmental conditions

You might also like

1. Introduction introduces a new set of challenges to this longstanding issue. Conse