Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Next Article in Journal
Exploring the Integration of Artificial Intelligence-Based ChatGPT into Mathematics Instruction: Perceptions, Challenges, and Implications for Educators
Next Article in Special Issue
Mobile Smartphones as Tools for ICT Integration in Geography Teaching
Previous Article in Journal
Acceptance of Educational Artificial Intelligence by Teachers and Its Relationship with Some Variables and Pedagogical Beliefs
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Systematic Review

Educational Approaches with AΙ in Primary School Settings: A Systematic Review of the Literature Available in Scopus

by
Spyridon Aravantinos
1,
Konstantinos Lavidas
1,*,
Iro Voulgari
2,
Stamatios Papadakis
2,*,
Thanassis Karalis
1 and
Vassilis Komis
1
1
Department of Educational Sciences and Early Childhood Education, University of Patras Greece, 26504 Rio, Greece
2
Department of Early Childhood Education, National and Kapodistrian University of Athens, 10680 Athens, Greece
*
Authors to whom correspondence should be addressed.
Educ. Sci. 2024, 14(7), 744; https://doi.org/10.3390/educsci14070744
Submission received: 5 June 2024 / Revised: 24 June 2024 / Accepted: 4 July 2024 / Published: 6 July 2024

Abstract

:
As artificial intelligence (AI) becomes increasingly prevalent, it has become a topic of interest in education. The use of AI in education poses complex issues, not only in terms of its impact on teaching and learning outcomes but also in terms of the ethical considerations regarding personal data and the individual needs of each student. Our study systematically analyzed empirical research on the use of AI in primary education, specifically for children aged 4–12 years old. We reviewed 35 articles indexed in SCOPUS, filtered them according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, analyzed them, and categorized the findings. The research focused on the studies’ objectives, learning content, learning outcomes, learning activities, and the pedagogy of activities or the AI tools. Our categorization resulted in three main categories of research objectives regarding the creation, implementation, and evaluation of AI tools and five categories for learning content: AI and ML (machine learning) concepts in STEM and STEAM, language learning, mathematics, arts, and various other subjects. The learning activities were split into four categories: apply, engage, interact, use; project-based learning with multiple activities; experience and practice; and students as tutors. The learning outcomes were split into three levels: cognitive, affective, and psychomotor. The pedagogy of AI tools falls into four categories: constructivism, experiential learning, AI-assisted learning, and project-based learning. The implications for teacher professional development are discussed.

1. Introduction

The term “artificial intelligence” (AI) can be defined in sufficient detail by encompassing its various aspects, which include the ability to achieve complex goals in complex environments, the explanation and simulation of intelligent behavior with computational processing in the context of rationality and logical reasoning, and the demonstration of behavior and intelligence that simulates human behavior and intelligence to achieve specific goals [1,2]. From the outset of AI, there has been a connection with the field of education that was driven by the desire of its pioneers to link AI to the learning process, to understand both how AI and learning work, as they emphasized the importance of cognitive science for the development of AI for educational purposes [3]. Nowadays, as the impact of AI increases and permeates more areas of our daily lives, the scientific field of education can not remain unaffected. Therefore, a rise in artificial intelligence in education (AIEd) research has been observed [4,5].
Additionally, over the last five years, some systematic reviews have tried to figure out the trends of AI and its educational implications. Specifically, Younis et al. [6] examined the applications of robots and natural language processing in education by studying 82 scientific articles from 2014 to 2023 with students of all ages. Even though their systematic review suggests that these technologies can provide feedback and personalized instruction, facilitate collaborative learning or critical thinking, and promote inclusivity, student engagement, and teacher support, they have not focused on learning content and activities or pedagogy of the AI applications. İpek et al. [7], in their systematic literature review of educational applications of ChatGPT with 40 studies from December 2022 to February 2023 and unspecified student ages, categorized ChatGPT’s use into positive and negative themes and focused on the implications, challenges, or potential effects of integrating ChatGPT into education and not on its pedagogical use or learning approaches. Positive use, relative to the education field, includes abstracting, literature review, generating literature, translation, and paraphrasing, generating complex and deep answers for exams, identifying students’ needs earlier, personalized learning experiences, grading and assessment, data analysis, prevention of cybercrime and cyberbullying, helping people to study, etc. On the other hand, potential problems with the use of ChatGPT can be cheating, creating bias, generating incorrect answers, and legal and ethical issues.
More related to our research aim are the following reviews. The systematic literature review on teaching and learning machine learning (ML) by Sanusi et al. [8] was not limited to a specific year range of research for its 43 articles, and even though about half included our target age of participants, it focused partly on pedagogy and not at all on learning activities or outcomes. The research explored ML teaching and learning in K-12 education from four development perspectives: curriculum, technology, pedagogical approaches, and professional development. The findings revealed that there needs to be more research on curriculum development, teacher professional development, and training in the context of ML. Additionally, there is a need for more ML resources for preschool and middle school levels and further evidence of the societal and ethical implications of ML.
Furthermore, Yim and Su [9], in their scoping review of AI learning tools in K-12 education with 46 studies from 1995 to 2023, more than half of which related to the preschool and primary school ages, focused on learning tools, learning outcomes, and the significance of innovative pedagogical strategies involved, but were mainly engaged to the context of AI literacy, exploring the way AI should be taught and not the way of teaching with AI. The review by Su et al. [10], which aimed to examine the thematic and content analysis of 16 empirical papers from 2016 to 2022, also identified challenges and opportunities emerging in digital literacy in early childhood education (ECE). However, although it addressed the learning content, pedagogies, and learning outcomes, it did not analyze the learning activities or consider students aged 6 to 12. The results point out challenges like teachers’ need for more AI knowledge, skills, and confidence and the lack of teaching guidelines or an appropriate curriculum design; on the other hand, they point out opportunities for students to enhance digital literacy skills and attitudes. Su and Yang [11] also conducted a scoping review. They analyzed 17 studies from 1995 to 2021 about AI in ECE regarding research methods, AI tools and knowledge, activities, and impacts on teaching and learning. The researchers demonstrated improvements in teaching and learning, but more research is needed on AI tutoring systems for younger students. Although the review focused on ECE, AI literacy, and learning activities, there was no reference for the learning content or learning outcomes. The literature review of 39 papers from 2018 to 2022 by Crescenzi-Lanna [12] about human–machine cooperation in education and its ethical implications was also concentrated solely on ECE; it noted that AI challenges are present in ECE in terms of data privacy, and it also examined other aspects of AI in educational settings such as data collecting and processing and predicting events related to students’ success and assessment, but it did not mention any pedagogical or learning aspect.
This work aims to present a systematic review providing valuable insights about the research trends for the educational applications of AI in preschool and primary school (age range 4 to 12). With this systematic review, we will cover the scarce presentations about the pedagogical approaches of AI in education in previous systematic literature reviews and the lack of empirical evidence for the students of our target age. Previous systematic literature reviews to date are limited and have either focused scarcely on different aspects of AI, involved a broader or narrower age range of students, or partially addressed the learning dimension, so we focused on filling this gap. A pedagogical approach to the integration of AI into educational environments can be the catalyst for redesigning the curriculum and transforming teacher education, as educators not only have positive attitudes about implementing AI in the classroom and its positive outcomes, but they are also willing to enhance their knowledge and skills with professional training [13]. Through adopting innovative pedagogical approaches with AI technology, it will be possible to establish new practices that optimize teaching processes and meet students’ individualized needs to assist them in acquiring essential skills like critical thinking and problem-solving [14]. With all this ongoing development of existing and the design of new tools, the educational system is on the verge of a significant shift that will affect future approaches [15]. Educational policymakers, educators, counselors, and other stakeholders such as parents, local communities, and students are affected by all these rapid changes, so the scientific community should be ready to provide the needed answers.

2. Research Objectives and Questions

We aimed to systematically review empirical articles related to the educational implementation of AI in preschool and primary education and to examine the latest trends in the research field. More specifically, we explored the SCOPUS-indexed literature on integrating AI into education settings for our age group with a pedagogical focus. The age range includes students from early childhood education (4 years old) to the upper grade of primary school (12 years old). This paper will focus on (a) the research objectives of the articles reviewed, (b) the learning content, (c) the learning activities, (d) the learning outcomes, and (e) the pedagogy of the activities or the AI tools used. The research objectives refer to what the authors of each study state about their aim for conducting the specific research, and the learning content refers to the specific lesson, course, or subject being taught with the aid of AI [11]. The learning activities refer to the activities that students engage in during the learning process, and the learning outcomes refer to what the students are learning regarding cognitive or non-cognitive skills [16]. The cognitive learning outcomes refer to specific skills and knowledge, whereas the non-cognitive learning outcomes include skills like cooperation, communication, critical thinking, problem-solving abilities, and motor skills [10]. The pedagogy of the activity or the AI tool is the one described in the studies for the design of the learning activities or the AI tools that are used in each case and can refer to various pedagogical approaches like direct instruction, inquiry-based learning, design-oriented learning, collaborative learning, interactive learning, project-based learning, hands-on activities, and participatory learning. To design an effective pedagogical model, elements like objectives, content, strategy, group organization, time and space allocation, selection of resources, evaluation, and feedback should be considered [17].
Our research questions were as follows:
RQ1: What are the objectives of the studies regarding the implementation of AI?
RQ2: What is the learning content of the teaching process with AI?
RQ3: What learning activities do students engage in each study with AI?
RQ4: What are the learning outcomes regarding cognitive and non-cognitive skills?
RQ5: What are the pedagogies of the activities or AI tools used?

3. Methods

For our research, we used SCOPUS as a single database with unrestricted access because of its content coverage, its convenience and practicality, and its more trustworthy impact indicators, which cannot be manipulated as easily as those provided by WOS [18]. The choice of SCOPUS has also been decided since it may offer better specialized coverage in this field of human–computer interaction, which is highly relevant to our review [19]. The string query for the search that occurred in January 2024 was the following:
TITLE-ABS ((preschool OR pre-school OR “primary school” OR “elementary school” OR “early years”) AND “artificial intelligence” AND (us* OR utiliz* OR implement* OR intervent*)) AND (EXCLUDE (SUBJAREA, “MEDI”) OR EXCLUDE (SUBJAREA, “HEAL”) OR EXCLUDE (SUBJAREA, “BUSI”) OR EXCLUDE (SUBJAREA, “BIOC”)) AND (LIMIT-TO ( DOCTYPE, “cp”) OR LIMIT-TO (DOCTYPE, “ar”) OR LIMIT-TO (DOCTYPE, “cr”)) AND (LIMIT-TO (LANGUAGE, “English”)).
Figure 1 presents the process of the method we have chosen for identifying and screening the articles, which follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic literature reviews [20]. For the screening process, we used only empirical studies and focused on educational applications of AI in school settings from early childhood to the highest grade of primary school. The index search yielded 194 entries, of which 3 were duplicates, 18 were excluded by title, 32 by abstract, 58 were unavailable, and 1 was unreachable. After removing the above, we ended up with 82 articles, of which 3 were previous systematic literature reviews, 3 were retracted, 2 were not full papers, 16 were theoretical, and 23 referred to technical terms about the computer systems used in AI for education, so they were not directly related. An overview of the final 35 empirical studies that met the inclusion criteria for the review can be found in Table A1.
Journal articles and conference papers of unspecified date were included, although most of the studies were published in the last three years, more specifically, 11 in 2023, 8 in 2022, and 7 in 2021. Papers in languages other than English or unrelated scientific fields such as biochemistry, genetics and molecular biology, business, management and accounting, health professions, and medicine were excluded. The reason we included articles from conferences as well as journals was to allow us to identify the most recent developments being addressed by the research community. The chosen studies had students of our target age as participants. However, we excluded studies with only educators or parents, including those with mixed groups, such as teachers and students (Table 1). Most research took place in China (N = 9), Taiwan (China) (N = 5), South Korea (N = 3), the USA (N = 2), Ecuador (N = 2), Indonesia (N = 2), and Japan (N = 2). Other studies were conducted in Germany, Israel, Singapore, India, the Russian Federation, Bangladesh, Finland, Slovenia, the Netherlands, and Sweden. Asia was the continent with the most research (N = 25), followed by Europe (N = 6), North America (N = 2), and South America (N = 2). Regarding the age of the participants, 27 of the empirical studies had primary school students with ages 6 to 12 as their target, 4 focused on preschool with ages 4 to 6, and another 4 had mixed ages of participant students.
All the essential data from the 35 eligible articles were extracted using an EXCEL spreadsheet with different themes, where the crucial information was recorded in the appropriate cells, as described by Crescenzi-Lanna [12]. When everything was coded into the spreadsheet, our next step was to look for similar patterns in relevant fields of the studies and perform a thematic analysis to highlight some significant themes by analyzing the data. More specifically, we used the ideas and themes we were investigating to make a coding scheme, which we applied to the data and identified relevant categories within our study fields. We then conducted the analysis where the themes we identified were examined, and we interpreted any connections or distinct patterns among them [21]. Regarding our research questions, the studied categories included (a) the research objectives, (b) the learning content, (c) the learning activity, (d) the learning outcomes, and (e) the pedagogy of the activity or the AI tool.

4. Results

The results of the previous process are presented below for each of the five themes of the studies we focused on for this review.

4.1. Research Objectives

Figure 2 presents the research objectives of the studies, which can be divided into three distinct categories. First and most often, researchers wanted to discover how integrating AI into education can enhance teaching and improve student learning outcomes. For this reason, new tools to support students were designed, developed, and tested, and existing ones were evaluated. The focus was on different cognitive areas, such as using music programs to enhance creativity [22], the improvement of handwriting with educational Robot Kiddo [23], and the promotion of computational thinking [24] by constructing a design-based STEM + AI teaching model. In addition to these, the development and evaluation of AI-based teaching and learning models like the CP3 in the context of converging multiple subjects were reported [25], and the design of an alternative teaching system for preschool education specialty courses to assist and guide educators to accurately and efficiently retrieve curriculum resources was tested [26]. There was also the development of robotic quiz games to promote self-regulated learning and increased learning engagement [27], in conjunction with AI and its role in promoting cognitive development and physical health of students through interactive learning with “Internet +” and “Big data ML” [28]. In this category, teaching material that integrates AI and ML concepts was constructed in the context of social and science education [29], as well as an ability-oriented STEAM-graded teaching system [30]. AI-based solutions to help identify dyslexia in primary school pupils [31] and AI applications as immersive learning environments that can assist children with Down syndrome [32] were developed. In the form of Natasha Bot, virtual learning partners were created for children with visual impairments [33], to improve their learning outcomes. Research also focused on the innovation of teaching methods and learning systems, with examples of voice assistants being used to assist children in using the toilet independently [34], and the development of intelligent educational and English teaching resources [35]. Completing the first category, we found a program (CAI) designed to assist students’ learning of open-sentence mathematical problems [36] and a recognition learning system for mathematical concepts such as natural numbers [37].
Moving to the second research category, we included studies that aimed to investigate the implementation of a tool as a teaching aid in terms of learning outcomes, effectiveness, or AI–student interaction. Podpečan [38] examined students’ engagement by investigating their emotional responses, and Williams et al. [39] explored their interactions with social robots to learn about AI. Wu and Yang [40] studied AI science activities in informal curricula on students’ AI achievement, and Chen et al. [41] used an AI-based children’s digital art ability training system to improve their imagination and drawing skills. The motivation to understand and learn ML through programming and computational concepts was accomplished in a constructivist learning environment and with the use of appropriate scaffolds, students managed to construct a neural network in the study of Shamir and Levin [42] which investigated course scaffolds and course outcomes.
Furthermore, personalized and adaptive learning experiences for English language learning and interactions were examined [43] by implementing an AI coach as a humanized agent following the CoI framework [44]. Educational games were also explored to further enhance the learning experiences offered and the engagement in mathematics [45], and Wu et al. [46] analyzed attitude, motivation, and cognitive load on continuous learning intention in STEAM education and learning outcomes associated with AI-assisted educational activities. The impact and the role of AI and robotics regarding teaching and learning practices were also addressed concerning physical development, social–emotional skills, and intellectual growth among students [47] and by identifying the pedagogical and technical considerations when designing teaching interventions with Google Teachable Machine (GTM) [48].
The last category of research was concerned with the evaluation, assessment, and analysis of the impact of AI on children’s learning and interactions. The influence of social and emotional intelligence on student character development in educational settings was assessed with the PKES instrument [49], and Neurofeedback technology with the aid of AI was used to analyze the level of attention and temperament of children [50]. Students’ perceptions and the process of accepting intelligent machines such as robots were tracked [51], and the influence of AIEd on adolescents’ social adaptability and emotional intelligence was studied [52]. Ethical implications were also brought into this category by testing the effectiveness of an AR-based contextualized dilemma discussion approach and by studying concepts such as trust, privacy, and the responsible use of technology [53], as well as an AI-automated analysis of digital storytelling for interdisciplinary learning, with integrated CT skills, for real-time adaptive feedback in STEM education [54]. Finally, Kajiwara et al. [55] examined the educational impact and changes in impressions of AI before and after role-playing the ML process by practicing problem-solving skills and computational thinking, and Bingi et al. [56] facilitated and evaluated learning with AI-based education tools and a humanoid.

4.2. Learning Content

The theme of learning content can be split into four main categories, which contain most of the studies, and a fifth one, which includes many studies with distinct learning contents, as shown in Figure 3.
The first category consists of STEM [24] and STEAM concepts [30,46] and AI and ML elements like mechanical expression, perception, reasoning, and consciousness, as well as their integration into educational environments [42,48,55]. Connecting such objects to real-life stories or problems and digital storytelling with incorporated physical science concepts [54], combined with scientific adventures mixing biology and AI concepts [29], can engage students in STEM education. The use of robotics and its applications in education, where robots like NAO bot or Natasha Bot were used as intelligent learning partners for students [38], can also be included here along with engineering, robotics courses, and other STEAM concepts. Learning AI concepts like knowledge-based systems, supervised machine learning, and generative AI [39], as well as sorting network algorithms and data [25], AI knowledge, coding, AI visual applications, and problem-solving through programming [40] or robotics [47], is a final branch of this category.
The second category refers to mathematics and geometry, where children learn basic arithmetic concepts and reasoning by doing and by acting as tutors for teachable agents [45] or are taught to recognize and write natural numbers from 0 to 10 [37]. In this category, students also formulated mathematical questions and added or subtracted numbers with a computer-assisted instruction (CAI) program [36]. Children with Down syndrome were assisted in recognizing or classifying geometric drawings [32]. The development of quiz games with fifth-grade math questions by Weng et al. [27], along with problem-solving abilities and a robot companion, was also found in this category.
The third category includes language learning with an AI coach for speaking and listening [43,44], where researchers were concerned with teaching vocabulary and all its aspects (word, sentence, meaning) in the context of English teaching as a second language. Huang [35] designed and developed educational robot teaching resources using AI to enhance English language teaching.
The fourth category is related to art education, where students can improve their painting performance [41] and enhance their understanding of music theory and harmony [22].
The rest of the studies were scattered among many learning contents, and one instance of each category was found. There is a study that explored ethical dilemmas and the role of educators in implementing AI technology [53] and other studies focusing on environmental issues [56], physical health education [28], or the promotion of children’s autonomy with independent toilet training [34]. Omokawa and Matsuura [51] focused on the theme “What is life for me”, exploring the moral notions of students about life by interacting with a humanoid, while Khilmiyah and Wiyono [49] explored the emotional and social responses of students with an instrument (PKES) that measured the learning outcomes. Andinia and Isnainiyah [33] accommodated a virtual learning partner to assist students with vision impairment disabilities answer questions from several subjects. Additionally, Mispa and Sojib [23] practiced handwriting and drawing skills with the aid of the robot Kiddo, and Shalileh et al. [31] proposed a robust AI-based solution to identify dyslexia in primary school pupils. Another category that we found is connected to curriculum development and teacher guidance and included topics related to information technology and programming in conjunction with AI specialty courses [26]. Finally, Lee et al. [50] proposed a system to predict and analyze children’s temperament and attention levels to investigate possibly hidden cognitive disorders, and Lai et al. [52] aimed to study the influence of AIEd on adolescents’ social adaptability through multiple courses from AI curriculum reform experimental schools.

4.3. Learning Activity

The learning activities students engaged in were categorized into four main groups, although they were often interconnected due to overlaps (see Figure 4).
The first category focuses on interactions with various platforms, tools, or agents, application engagement, participation, and implementation. Thus, Choi [22] examined the music program Doodle Bach to assist students in creating music compositions, Shamir and Levin [42] had students construct an ML-based artifact using a programmable environment, and Barnard et al. [36] used a computer-assisted program (CAI) for students to diagnose problem-solving strategies and their misconceptions. In the study of Gupta et al. [54], students used a learning environment to engage in problem-solving activities and the creation of interactive science narratives, and in the study of Napierala et al. [29], students played with a memory game to identify leaf types in biology, testing decision trees in AI and ML concepts at the same time. There was also an interaction with a virtual environment for children with Down syndrome, using the Leap Motion Controller for kinesthetic engagement [32], and an interaction with a voice agent to guide and support the toilet training process of young children [34]. Bingi et al. [56] referred to interactive learning activities like listening, answering, and receiving feedback from the humanoid NAO, and Wang et al. [43] described an AI coach for EFL learning that aids students in practicing speaking and listening, as well as pronunciation [44]. Additionally, students engaged with the PopBots platform to answer multiple-choice questions [39] and used applications for question-and-answer sessions with Natasha Bot [33]. Participation in math quizzes with the AI robot Zenbo by Weng et al. [27] and writing on a shared whiteboard with the robot Kiddo [23] were also included in this category.
Moving to the second category, we find studies incorporating multiple activities in the form of project-based learning, to provide students with holistic approaches suited to various needs and to enhance their learning outcomes. Here, there were real problem situations that promoted active participation in solving them with peers, and programming and model-building activities for improving computational thinking [30]. In the study of Li et al. [24], students focused on creating scenarios that contained practical problem-solving by designing and implementing solutions to promote scientific literacy and interdisciplinary exploration. In the study of Huang [35], students engaged in role-playing scenario creation, questioning and answering sessions, classroom interactions, group discussions, cooperative gaming, and knowledge carding. In the study of Wu and Yang [40], students were involved in a 6 h AI education program based on the STEM learning conceptual framework and project-based learning, which included teaching sessions, hands-on exercises, group problem-solving activities, and designing and solving problems in real-life scenarios. Podpečan [38] demonstrated multiple applications: motor development activities and games, children’s games, theatrical performances, artificial intelligence applications, and data harvesting applications, based on the main topic and programming techniques for the NAO robot. Joo and Park [25] engaged students with unplugged hands-on activities and natural interaction exercises, for tool exploration and understanding the social impact of AI, as well as with games for the collection and analysis of data, targeting the enhancement of problem-solving and reasoning skills. Students also used web-based tools for an introduction to ML fundamentals, participated in group discussions to exercise critical thinking and creativity, and were assigned a project to work on with Google Teachable Machine (GTM) in the work of Toivonen et al. [48], with the aim of training models and developing applications to conclusively reflect and obtain feedback for their actions. Finally, Wu et al. [46] integrated STEAM with AI into their activities to provide children with hands-on opportunities to create intelligent systems, understand vision recognition, code a program, assemble a game, make graphs, and evaluate and tune correct parameters.
The third category consists of more experiential and practicing activities that can enhance expression and creativity, like observing simulations and role-playing for an inquiry, based on AI issues and ethics in real life [53]. There was also the experience of a machine learning role-playing game (ML-RPG) where students engaged in tasks related to the ML process [55]. In the work of Salas-Pilco [47], students brainstormed, designed advanced robotic models to solve community problems, and presented their solutions, demonstrating social responsibility. Moreover, students collaborated with the NAO robot and its programmer, pretended to be humanoid, watched a movie about care robots, or wrote reflectively in the study of Omokawa and Matsuura [51]. The practice of art painting finally included a series of activities like the creation of the work, the recognition of its technical details, and the experience of an augmented reality (AR)-enriched exhibition of paintings [41].
Lastly, we have two cases where students acted as the tutors for the teachable agent game or the system, interacting with it by playing and answering situation-specific questions [45] or practicing mathematics [37].

4.4. Learning Outcomes

Regarding the learning outcomes, we categorized them into three levels: cognitive skills, affective skills, and psychomotor skills [57], with the focus of the literature being on the affective level. Adopting new desirable behaviors and critical thinking, knowledge acquisition, an overall increase in learning, and presenting a higher quality of work from students are direct observations from the data synthesis. Figure 5 shows the frequency of reports about learning outcomes per level, and Figure 6 presents a word cloud of all learning outcomes.
The first category of affective skills includes reports of social and emotional development of the participants [56], as there was an acknowledgment, appreciation, and management of their emotions [49], as well as an increase in social responsibility and commitment [47], and cooperation with and respect for others in social interactions [30]. Emotional engagement with robotic assistants, empathy for them, and higher levels of learning satisfaction and interest were also recorded [51]. In general, self-efficacy and self-confidence increased as well [47,49], and a shift in students’ attitudes to more favorable levels with less anxiety and less fear of accepting AI was observed [55]. In this category, we included autonomous learning [30,35], responsibility, communication [29], collaboration, and increased reports of student engagement [25,37,39,45,46], in line with students’ overall improvement in a variety of areas. Students’ interest and enjoyment in learning increased, as did their reflective ability and intrinsic and extrinsic motivation [37,41,43,44,46], enhanced attitude [46], and the acquisition of new behaviors with behavioral shaping techniques [34].
Regarding the cognitive level, there were learning outcomes about understanding AI [42], understanding core machine learning concepts and how students use this practical knowledge to train ML models or build applications [48], and understanding computational thinking [25]. Understanding the ethical issues that arise from the use of AI in an educational context [53] was one of the learning outcomes, as were the discussions within the specific bounded context of the possibilities and limitations of technology, especially regarding the perception of robots [39]. The recognition of numbers and geometric figures and the comprehension of the quantities represented by numbers was another learning outcome [32,37], along with proficiency in science concepts and story structuring [54]. With the development of student’s ability to solve a problem [34] came the acceptance of AI [55] and insight into processes [36] requiring improved competence and the development of technical skills [46]. Improvement was also observed in many areas, like work quality [40], pronunciation, listening comprehension and vocabulary [44], artistic expression, originality, performance, and creativity [41], music cognitive abilities [22], learning new expressing abilities [24], and technology application ability [27].
At the psychomotor level, there was an improvement reported in the fine motor skills of students, with visual recognition and categorization of geometric figures in mathematics [32], and motor mimicry involving handwriting and drawing skills [23].

4.5. Pedagogy of Activity or AI Tool

Figure 7 shows the pedagogies used in the studies, which we split into four main categories. The pedagogies used were usually mixed, so most studies belong to more than one category.
Among the pedagogical methods used, the most common was experiential learning, with activities adapted to the student’s interests, abilities, and experiences and with the employment of immersive environments [32]. Additionally, here, we included problem-solving activities [30,33,43,47], with the exercise of computational thinking skills [25,54], hands-on learning through interactive objects [29,40,47], and interactive narratives [54], as well as role-playing games or scenarios [35,53,55]. In this category, the transfer of knowledge and learning [45,53] and promotion of ethical reasoning skills [53] were also found as means of learning, along with the interaction between students, teachers, and AI [56].
In the second category, technology was used to enhance pedagogical methods. Here, we found social robots as learning assistants [38], adapted to support self-efficacy and self-regulation of learning [27,49], or even taking the role of the learner, positioning the student as the tutor [37,45]. Gamification elements strongly contributed to learning engagement and increased motivation, providing tutoring guidance and feedback [37], and auxiliary teaching systems were used to overcome deficiencies of traditional teaching and enhance shortcomings of existing courses [26]. Collaborative hands-on learning by teaching was also found [23], along with interactive dialogues with AI agents [34]. The involvement of AI in the learning process enhanced interest and upgraded the quality of the material provided [22], emphasized motivation, guided observation and thinking, and encouraged creation [41], personalizing the feedback that was given by the virtual intelligent teacher to each individual learner [44].
In the third category, we found project-based learning, defined by the objectives pursued and implemented by the project method for acquiring STEAM skills [30]. The activities were child-centered, and the research was carried out through project work [47] to resolve the problem and find innovative solutions. Participatory and active learning [39], as well as situational assessment [56], placed the students in truth-seeking scenarios of inquiry [46,53], with the guidance of the AI tools in question. There was also a combination of activities, including lectures, exercises, and group discussions, to come up with a solution [40], or to promote logical reasoning skills on the ethical issues of AI [53], as well as action-oriented learning and design-based research [29].
Lastly, constructivism was another emerging category that stood out as a theory that supports the active involvement of students in the construction of knowledge [24,42,51]. Group collaboration, in conjunction with critical thinking activities, contributed to the acquisition of new knowledge [39], which was adapted to the needs and capabilities of the students [36]. Children became the designers and constructors of their own learning [48], and by reflecting on AI [39], and prompting for self-explanation [45], students managed to refute any misconceptions they may have had.

5. Discussion and Implications

5.1. Relevance with Previous Systematic Reviews

This paper provided insights regarding the research objectives of the relevant studies, which we divided into three main categories according to the researcher’s approach to the AI tools. Most of the studies proposed and designed new tools, others implemented and explored their applications, and the least of the studies evaluated their impact on teaching and learning outcomes and students’ responses. Some similarities occurred with the systematic review of Sanusi et al. [8] in learning aims, e.g., AI concepts, social robots, games, understanding of ML, and decision trees, and with Su and Yang [11] in assessing the effectiveness of platforms and robots, exploring multiple applications, and investigating their use, children’ s perceptions, and learning outcomes.
The learning content of our studies was mainly focused on AI and ML concepts like the learning content reported in [10], e.g., experience machine learning, knowledge-based systems, AI robots, and ethics. Our findings were also focused on language and math learning; however, some of our studies were spread throughout many different subjects, e.g., AI ethics [53], physical health promotion [28], environmental awareness [54,56], children’s autonomy [34], moral notions about humanoids [51], emotional and social responses [49], handwriting and drawing [23], learning partner for vision impairment disabilities [33], dyslexia identification [31], specialty courses [26], temperament and attention levels of children [50], curriculum reform [23], and social adaptability [52].
Regarding the theme of learning activities and more specifically the category where students practice and experience simulations, our findings had similarities with those stated in [8] about discussions, role-play, robot and simulation games, scientific inquiry, the use of Google Teachable Machine (GTM), and unplugged activities. GTM was also found in the reviews of Yim and Su [9], Su et al. [10], and Su and Yang [11], along with hands-on activities in [10], for assisting preschool students in their learning of concepts about knowledge-based systems and supervised machine learning. Su and Yang [11] also mentioned the effectiveness of intelligent tutoring systems, problem-solving, peer-to-peer interactions, and creative inquiry literacy, as we also did.
Cognitive learning outcomes like understanding of AI, machine learning, or knowledge-based systems were also reported by Yim and Su [9] and Su et al. [10], along with affective and behavioral outcomes, e.g., motivation, self-efficacy, high student engagement, collaboration, and communication skills. Affective skills were reported more than any other category in our study, in contrast to the review of Su et al. [10] in which only two researchers had designed activities to enhance students’ higher-order thinking skills. On the contrary, creative, emotional, and collaborative inquiry were reported as improved skills, although there was no mention of attitudes, motivation, and confidence [10].
Our findings on the theme of pedagogy showed that project-based learning, ML-based solutions, active-based and participatory learning, collaborative methods, and lectures were utilized, as was found in [8]. The most popular pedagogy used in our studies was experiential learning, which in combination with project-based and constructivist methods as interrelated theories, comes in agreement with the findings of the review of Yim and Su [9]. Human–computer interaction and child-centered play-based active learning follow, while the findings in [10] are similar to ours regarding activity-based and experiential learning. Project-based approaches are used less in preschool than in primary schools, while creative inquiry literacy is more suitable for younger students.

5.2. Teachers’ Role and Skills

The use of AI and its implementations in preschool and primary education (ages 4 to 12) is an area of growing interest and has experienced significant growth in recent years. Integrating AI in school environments should promote empowering experiences for students and support teachers to further enhance the quality of education. Technology and AI tools aid the teaching process by enriching the learning content and can enhance educators’ work, improving students’ knowledge and skills. Teachers play a crucial role in facilitating student interactions with AI, guiding and evaluating their use of technology in creative ways. For this reason, educators need to prepare themselves to effectively utilize new AI tools and integrate them into classroom settings to improve learning outcomes [58]. The role, required skills, and perceptions of educators, as well as teacher training for improving knowledge, are expected to be of significant concern for the educational community in the immediate future [59].
Moreover, the development of an AI literacy-implemented curriculum should be organized based on teachers’ perspectives and students’ needs and provide the appropriate tools to promote learning [60]. On the other hand, the issues raised by the implementation of AI in educational settings are complex, both in terms of its impact on teaching processes and learning outcomes and the ethical dilemmas involved [61]. All stakeholders must be aware of these rapid advancements, and educational policymakers must focus on organizing teacher professional development opportunities and on incorporating AI approaches into curriculum design.

5.3. Suggestions

Considering the importance of teachers for education generally and more specifically for facilitating the interactions of students with AI, educators’ AI readiness [62] is vital for the implementation of AI in school settings. Therefore, we suggest the following:
  • Focus on teacher training programs and professional development of educators, according to their specific needs. Some topics of interest could be as follows:
    AI-assisted applications for various everyday class activities and routines;
    Artificial intelligence, computational thinking, and machine learning basics;
    Artificial intelligence implications and challenges like AI ethics and inclusivity;
    AI implementations of platforms and tools, teaching methods and assessment, and curriculum design;
    Teachers’ training should be remunerated, take place during working hours or on educational leave, to surpass any resistance or difficulties and enhance their positive view of technology, ensuring the ethical implementation of AI in preschool and primary school education.
  • Further research is needed on implementing AI in preschool (ages 4 to 6) mostly, and primary school (ages 6 to 12), for teaching multiple subjects and addressing students’ individual needs.
  • Research on implementing AI on various courses and student ages, e.g., adult education, musical instrument courses, AI literacy, history, differentiated learning approaches, and personalized feedback for the tutor and the learner.
  • Examine the effectiveness of AI implementation in terms of pedagogical strategies and learning outcomes according to the student’s age.
  • A theoretical framework and/or policy guidelines for successful AI implementation in educational settings could be developed so that teachers can rely on it to increase the adoption of such technologies in their classrooms.

5.4. Limitations and Future Recommendations

The first limitation of this paper is that the articles that were not accessible at the time of retrieval might contain critical empirical evidence needed for this review. Another shortcoming of this systematic review is that it did not consider other databases and indexes, such as WOS, and it used only literature from SCOPUS. Also, a noted consideration for our research is that the themes we studied are inextricably linked, making it difficult at times to differentiate our findings, due to the size of overlapping information.
Future research could focus on preschool education to provide a more detailed overview of AI approaches and their learning outcomes, or other implications with students of this age. Even though the studies included in this review were disproportionally more about primary school than those focusing on preschool, further research is recommended for all school ages, as we think that the AI curriculum development should be unified for all levels of education. Research should also occur in more countries, and within different educational contexts, as most of the empirical studies that we found were located in Asia, whereas continents like North and South America or Europe had the fewest.

Author Contributions

Conceptualization, S.A., K.L. and I.V.; methodology S.A., K.L. and I.V.; software, S.A., I.V. and S.P.; validation, S.A., K.L. and I.V.; formal analysis, S.A., K.L. and S.P.; investigation, S.A.; resources, V.K.; data curation, T.K.; writing—original draft preparation, S.A, K.L. and S.P.; writing—review and editing, S.A., K.L. and S.P.; visualization, I.V.; supervision, T.K. and V.K.; project administration, T.K. and V.K. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on reasonable request.

Acknowledgments

This paper has been supported by the funding program “MEDICUS” of the University of Patras.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Table A1. Summary table of all studies and areas of focus.
Table A1. Summary table of all studies and areas of focus.
No.ReferenceCountry/
Region
Research AimLearning
Content
Learning ActivityLearning
Outcomes
Pedagogy of
Activity or AI Tool
1Andinia and Isnainiyah (2020) [33]IndonesiaVirtual learning partner Natasha Bot for people
with vision
impairment
disabilities
Questions
about subjects
in schools,
presented in
the form
of guessing
or trivia
Use the
application and
engage in
a question-and-
answer
processing
session with
Natasha Bot
Addressing
psychological
aspects of
learning for
individuals
with disabilities
Design thinking approach,
problem-solving, and user-centered design to create
innovative
solutions
2Barnard et al. (1988) [36]NetherlandsDevelopment
of a computer-assisted instruction (CAI) program for open sentence
mathematical
problems
Elementary
mathematics,
open sentences
in + and -, the
identity of the
unknown, and
the operation
sign
Use the CAI
Program to
diagnose
students’
problem-solving strategies and
misconceptions
Gain insight into problem-solving processes
and improve
competence in solving such
problems
Adapt instruction to the level of
individuals
by diagnosing
their existing knowledge and
misconceptions
3Bingi et al. (2021) [56]IndiaFacilitate
and evaluate
learning
among students
by using AI-based education tools
and humanoid
Stories from NCERT
textbooks and
environmental
science topics
from Wikipedia
Interactive
learning
activities with
the humanoid
NAO (listening,
answering,
receiving
feedback)
Quality of social
interaction; warmth,
competence,
discomfort,
emotional
response, feelings for robot
Active learning and assessment,
interactive
and engaging
learning
experiences
4Chauca et al. (2023) [32]EcuadorDevelop two
applications as
immersive
environments
for children with Down syndrome
Geometric figures recognition and
classification and performing and
recognizing
numbers from
1 to 10
Interact with
the virtual
environment
using the Leap
Motion Controller to engage
with the
applications
Recognition and
classification of
geometric figures and numbers.
Improvement of fine motor skills
Immersive
environments,
interactive
learning
experiences,
supervised
learning
5Chen et al. (2022) [41]TaiwanImprove
imagination and drawing ability
using a children’s digital art ability training system based on AI
Students’
cognition of
chromatics and
enhancement
of students’
imagination
and painting
performance
Create work,
recognize outline, match hue color, calculate color
ratio, view
actual AR
paintings
Improvement
in originality,
flexibility, title
abstractness, and total scores,
imagination and painting
performance
Emphasizing
motivation,
guide
observation
and thinking,
encourage
creation
6Choi (2023) [22]South KoreaDesign,
implementation,
and effects of an
elementary music creation class
using AI-based music program
Tonality cognition, rhythm cognition, and melody
cognition
through the
program
Use AI-based
music program, Doodle Bach,
to create music compositions
Improvements in music cognitive abilities, growth in ability to perceive rhythm, positive impact of sharing creative works and providing feedbackIncorporated AI
as an active
‘media’ in the
lesson to engage students and
maintain
interest
in learning
7Gupta et al. (2023) [54]United StatesDigital
storytelling for
interdisciplinary learning,
science narratives that integrate
CT skills
Physical science concepts
and energy
conversions, aligned with
US science
standards
Use learning
environment,
engage in
problem-solving activities and create interactive
science
narratives
Proficiency in
science concepts, CT, and story structure.
Storytelling and
problem-solving strategies
Problem-solving scenarios,
interactive
narratives,
storytelling,
physical science, and computational thinking
8Huang (2021) [35]ChinaDesign and
develop
educational
robot teaching
resources using
AI to enhance English teaching
Vocabulary
teaching
functions.
Speech, meaning, example sentence, and unit of a
word input
by the user
Role-playing scenarios creation, vocabulary
teaching, classroom interaction,
Q and A, group discussions,
cooperative games, word detection and consolidation, teaching review and knowledge carding
N/AInnovative
teaching,
autonomous
learning,
interactive
and dynamic
learning
experiences
9Huh et al. (2022) [34]South KoreaDevelop a
service design
using AI voice agents to assist
in independent
toilet training
Using
the toilet
independently
Interact with an
AI voice agent named
Ddongddong
to guide and
support toilet training process
Developed
problem-solving abilities and
acquisition of new behaviors with
behavioral shaping techniques
Interactive
dialogues with
the AI agent.
The agent acts as
a practical tool
to provide
reinforcement
10Joo and Park (2022) [25]South KoreaDevelopment
and application
of an AI-based
convergence
education
teaching–learning model, the CP3
Sorting network
algorithm,
procedural
thinking,
sequential
structures, and procedural
thinking
Implementing
the CP3 model, which consists of problem
recognition,
planning,
and play stages
Computational thinking ability, higher
satisfaction,
interest, and
engagement
in the AI-based classes
Problem-solving with step-by-step computational thinking skills. Recognition,
planning, and
play stages
11Kajiwara et al. (2023) [55]JapanEducational
impact and changes in
impressions of AI before and after role-playing of the ML process
ML process,
AI decision criteria, math of ML, decision tree
models, and
classification
results
Experience a
machine learning role-playing game (ML-RPG) where they engage in tasks related to the ML process.
Understand ML, skills in perceiving, expressing,
reasoning and
learning,
self-efficacy and
acceptance of AI
Hands-on
experiential
learning through role-playing and
interactive tasks
12Khilmiyah and Wiyono (2023) [49]IndonesiaEffectiveness
of an android-based emotional
and social
intelligence
assessment
instrument
(PKES)
Emotional and
social intelligence aspects,
through cognitive, affective, and
psychomotor
domains
N/ARecognize,
appreciate, manage self-emotions.
Social
responsibility
and cooperation,
respect and tolerate others
Use of
experimental
methods
tailored to
student interests, abilities, and
learning
experiences
13Lai et al. (2023) [52]ChinaIdentify the
influence of AIEd on adolescents’
social
adaptability via
social support
IT, general
technology and
programming courses, flat panel teaching,
intelligent reading, assembling robots, 3D printing, Lego plug-ins, teaching boxes
N/AN/AN/A
14Lee et al. (2019) [50]TaiwanPredict and
analyze the
attention levels
of children
aged 4–7 years old
N/AN/AN/AN/A
15Li et al. (2023) [24]ChinaConstruction of a design-based STEM + AI
teaching model to cultivate
computational thinking
AI robot courses (voice, text,
automatic
translation,
companion, police, shopping guide, and
accounting robot)
Focus on the
creation of
scenarios that lead to the design of tasks related
to the intelligent learning partners
Ability to express, ability to question and ability to
connect in the
context of
computational thinking
New knowledge, question asking, collaboration, model building, share and display evaluation
feedback
16Lin et al. (2023) [53]ChinaAR-based
contextualized
dilemma
discussion
approach to
foster students’
AI ethics and
behavior
AI ethical
dilemmas
Observe
simulations,
inquire about
ethics, play
different roles,
explore AI issues
in real-life
contexts
using AR
Understanding ability for
complex AI
ethical issues.
Contextualized
discussing and
social interaction
Scenario
simulations,
in-depth inquiry, transfer learning stages, promote ethical reasoning skills
17Ma et al. (2021) [28]ChinaDesign and
implement a new teaching system for physical health promotion
with
“Internet +” and
“Big data ML”
A user interest model based on
DL algorithms,
an optimization
for health
promotion model teaching
N/AN/AN/A
18Mispa and Sojib (2020) [23]BangladeshRobot Kiddo for
Interactive
Handwriting
scenarios by providing
a shared
environment
for writing
100 basic shapes from elementary, grades 1 and 2
textbooks of the National
Curriculum
and Textbook
Children and Kiddo write
simultaneously
on a shared
whiteboard
Handwriting
and drawing
skills.
Motor mimicry
and cognitive
development
Interactive
handwriting,
collaborative hands-on learning by teaching,
playful and
engaging learning
19Napierala et al. (2023) [29]GermanyDevelop and
test teaching
material that
integrates AI
and ML
concepts
AI, ML, and
decision trees
in computer
science, the
structure and
features of leaves in the biology
section
Memory game
to identify leaf types, create
and test
decision trees based on leaf
features and
unknown leaves
AI-ML,
leaf types,
decision-making. Communication, biological terms, interactive work, connecting to
mathematics
Action-oriented learning and
design-based
research,
hands-on
activities,
discussions,
and reflections
20Omokawa and Matsuura (2018) [51]JapanDevelopment
of student
notions about
life from
their dialogs
with the
humanoid
robot NAO
Focus on
the theme of
“What is life for me?”
Collaborative
discussions,
watch a movie about a care
robot, pretend
to be NAO,
individual
reflective writing
Interest in NAO’s mechanical
functions.
Empathy and
emotional
connections with NAO
Constructivist
educational method,
interactive and
experiential
learning
21Pareto (2014) [45]SwedenTeachable agent and engagement
in math,
mathematical
skills and
performance
Basic arithmetic understanding,
the base-10
number system, fundamental
mathematical
concepts
Play the teachable agent game and
become a tutor to teach the agent,
answering
situation-specific questions
In-game knowledge to
traditional
mathematics,
engagement,
reflection, and
explanation
Reflect on
decisions, prompt self-explanation, support the
transfer of knowledge, and provide a role model
22Podpečan (2023) [38]SloveniaPhysical
embodiment,
anthropomorphism and the emotional aspects in
child–robot social interaction
Use of robotics,
engineering,
and artificial
intelligence to
engage students
in STEM
Develop and demonstrate
applications based on the main topic and programming techniques for NAO robot
N/AIntegration of
social robots
into education
and tutoring
23Salas-Pilco (2020) [47]ChinaAI and robotics
Impact
on learning
and teaching
activities, physical, social–emotional, and intellectual
AI and robotics technologies
to design and
create advanced
robotic models to solve community
problems
Brainstorm
solutions, select a key problem,
develop a robotic project to
address it,
and present
solutions
Imagining,
devising, testing. Self-confidence, teaching,
committing,
social
responsibility,
and presenting
Design-based
research,
integrated
analytical
framework,
hands-on
learning,
problem-solving
24Shalileh et al. (2023) [31]Russian
Federation
Propose a robust AI-based solution to identify
dyslexia in
primary school
pupils
N/AN/AN/AN/A
25Shamir and Levin (2021) [42]IsraelCourse scaffolds and course
outcomes in terms of motivation to learn and
understanding
ML
Machine learning (ML) and the
‘machine learning process’
Students construct an ML-based artifact
using a novel
programmable learning
environment
(PLE)
Increase in
students’
understanding
of AI concepts
and the ML
process
Constructionist learning method
26Shi and Rao (2022) [30]ChinaPropose and
realize a novel
ability-oriented STEAM graded teaching system
for high-quality teaching
Cultivation of
diversified abilities with the
development of cognitive and
non-cognitive
skills
Project-based learning in real problem situations, cultivating
abilities to solve practical problems
Autonomous
learning, problem-solving, critical thinking.
Responsibility, communication, cooperation
Reverse design based on the
ability goal,
project-based learning methods to achieve the
desired STEAM abilities
27Toivonen et al. (2020) [48]FinlandInvestigate the technical and
pedagogical
feasibility of Google Teachable Machine
Machine learning principles and
design of
ML-powered
Applications
using Google Teachable Machine
Conduct co-design workshops,
innovate and design ML-powered
applications with Google Teachable Machine
Understanding core ML concepts and practical knowledge for training an ML model and build applicationsChildren as
designers and
creators in
learning process
28Villegas-Ch. et al. (2022) [37]EcuadorDesign and
create an image
recognition
system to learn
natural numbers between 0 and 9
Recognition
and writing of natural numbers 0–9
The child acts as the tutor for the system, interacting with it to practice the numbersRecognize, write, and understand numbers,
comprehend the quantities
associated with each number
Gamification for motivation and
engagement,
system as a
teaching aid and tutor for guidance and feedback
29Wang et al. (2022) [44]SingaporeAI coach,
developed
for EFL learning,
can support
language learning following the CoI framework
English as a
foreign
language (EFL)
Students listen
to sentences read by the AI coach,
repeat them, and receive feedback
on their
pronunciation
Improved
pronunciation,
listening
comprehension, vocabulary.
L2 enjoyment,
affection for AI
Personalized
feedback, virtual intelligent teacher
30Wang et al. (2023) [43]ChinaCluster and
epistemic
network analysis
to provide
insights for
interaction with
AI coach for EFL learning
English as a
foreign language (EFL), improve speaking and
listening skills,
vocabulary
learning
Interact with
the AI coach for EFL learning,
practice speaking and listening,
as well as
vocabulary
learning
Deep, surface and
organized
approach to
learning.
L2 learning
enjoyment,
intrinsic and
extrinsic
motivation
Feedback,
problem-solving, agentic
exploration,
different
approaches,
motivation
31Weiwei (2022) [26]ChinaDesign an
auxiliary
teaching system
for preschool
education
specialty courses based on AI
PE course
resources,
curriculum
information and teaching
guidance
function modules
N/AN/AUsing AI
technology to
improve the
shortcomings of
existing PE
courses and
enhance the
teaching quality
32Weng et al. (2020) [27]TaiwanDevelop robotic quiz games for
self-regulated learning
Mathematics,
specifically
designing math questions for the program and
reviewing knowledge
learned in class
Participate in
math quiz games with the AI robot, Zenbo, to review and practice math concepts
Improvement of technology
application
ability,
enhancement of problem-solving, increase
in learning
Integration of
educational robots to enhance
learning of
programming
and support
self-regulated learning
33Williams et al. (2019) [39]United StatesInteraction with
social robots to learn AI
Knowledge-based systems,
supervised ML,
algorithms’ basic functionality,
edge cases and
initialization
Children engage with the PopBots platform and
answer multiple-choice questions
Understanding of AI, prediction and adjustment,
perception of
robots’ autonomy and limitations
Engagement in learning and
empowerment to reflect on AI.
Participatory
learning and
critical thinking
34Wu and Yang (2022) [40]TaiwanAI science activities in informal
curricula on
students’ AI achievement in popular AI
science activities
AI knowledge,
coding, AI visual recognition chip applications, and problem-solving through
programming
AI education
activity based
on the STEM
learning
conceptual
framework and project-based learning
Enhanced learning results and
creativity,
work quality,
computational thinking and
problem-solving skills
Combination of lectures, hands-on exercises, group problem-solving activities
35Wu et al. (2022) [46] TaiwanAttitude,
motivation, and cognitive load
on continuous learning intention in STEAM
education
STEAM (Science, Technology,
Engineering, Arts, Mathematics)
concepts and AI concepts
Design an
AI-based STEAM game that uses computer vision and controls a
robot to play a game
Development of technical skills.
Enhanced attitude, motivation, and continuous
learning intention in STEAM
STEAM
engagement,
exploration,
explanation,
engineering,
enrichment, and evaluation

References

  1. Athanassopoulos, S.; Manoli, P.; Gouvi, M.; Lavidas, K.; Komis, V. The Use of ChatGPT as a Learning Tool to Improve Foreign Language Writing in a Multilingual and Multicultural Classroom. Adv. Mobile Learn. Educ. Res. 2023, 3, 818–824. [Google Scholar] [CrossRef]
  2. Russell, S.J.; Norvig, P. Artificial Intelligence: A Modern Approach, 3rd ed.; Global edition; Prentice Hall Series in Artificial Intelligence; Pearson: Upper Saddle, NJ, USA, 2009. [Google Scholar]
  3. Doroudi, S. The Intertwined Histories of Artificial Intelligence and Education. Int. J. Artif. Intell. Educ. 2023, 33, 885–928. [Google Scholar] [CrossRef]
  4. Lavidas, K.; Voulgari, I.; Papadakis, S.; Athanassopoulos, S.; Anastasiou, A.; Filippidi, A.; Komis, V.; Karacapilidis, N. Determinants of Humanities and Social Sciences Students’ Intentions to Use Artificial Intelligence Applications for Academic Purposes. Information 2024, 15, 314. [Google Scholar] [CrossRef]
  5. Ye, R.; Sun, F.; Li, J. Artificial Intelligence in Education: Origin, Development and Rise. In Intelligent Robotics and Applications; Liu, X.-J., Nie, Z., Yu, J., Xie, F., Song, R., Eds.; Lecture Notes in Computer Science; Springer International Publishing: Cham, Switzerland, 2021; Volume 13016, pp. 545–553. [Google Scholar] [CrossRef]
  6. Younis, H.A.; Ruhaiyem, N.I.R.; Ghaban, W.; Gazem, N.A.; Nasser, M. A Systematic Literature Review on the Applications of Robots and Natural Language Processing in Education. Electronics 2023, 12, 2864. [Google Scholar] [CrossRef]
  7. İpek, Z.H.; Gözüm, A.İ.C.; Papadakis, S.; Kallogiannakis, M. Educational Applications of the ChatGPT AI System: A Systematic Review Research. Educ. Process: Int. J. 2023, 12, 26–55. [Google Scholar] [CrossRef]
  8. Sanusi, I.T.; Sunday, K.; Oyelere, S.S.; Suhonen, J.; Vartiainen, H.; Tukiainen, M. Learning Machine Learning with Young Children: Exploring Informal Settings in an African Context. Comput. Sci. Educ. 2023, 34, 161–192. [Google Scholar] [CrossRef]
  9. Yim, I.H.Y.; Su, J. Artificial Intelligence (AI) Learning Tools in K-12 Education: A Scoping Review. J. Comput. Educ. 2024, 1–39. [Google Scholar] [CrossRef]
  10. Su, J.; Ng, D.T.K.; Chu, S.K.W. Artificial Intelligence (AI) Literacy in Early Childhood Education: The Challenges and Opportunities. Comput. Educ. Artif. Intell. 2023, 4, 100124. [Google Scholar] [CrossRef]
  11. Su, J.; Yang, W. Artificial Intelligence in Early Childhood Education: A Scoping Review. Comput. Educ. Artif. Intell. 2022, 3, 100049. [Google Scholar] [CrossRef]
  12. Crescenzi-Lanna, L. Literature Review of the Reciprocal Value of Artificial and Human Intelligence in Early Childhood Education. J. Res. Technol. Educ. 2023, 55, 21–33. [Google Scholar] [CrossRef]
  13. Han, X. How Does AI Engage in Education? A Quantitative Research on AI Curriculum and Instruction in Public Primary Schools. In Proceedings of the 2021 4th International Conference on Education Technology Management, Tokyo, Japan, 17–19 December 2021; ACM: Tokyo, Japan, 2021; pp. 15–19. [Google Scholar] [CrossRef]
  14. Benvenuti, M.; Cangelosi, A.; Weinberger, A.; Mazzoni, E.; Benassi, M.; Barbaresi, M.; Orsoni, M. Artificial Intelligence and Human Behavioral Development: A Perspective on New Skills and Competences Acquisition for the Educational Context. Comput. Hum. Behav. 2023, 148, 107903. [Google Scholar] [CrossRef]
  15. OECD. The Future of Education and Skills: Education 2030 Position Paper. 2018. Available online: https://observatorioeducacion.org/sites/default/files/oecd-education-2030-position-paper.pdf (accessed on 22 June 2024).
  16. Druga, S.; Vu, S.T.; Likhith, E.; Qiu, T. Inclusive AI Literacy for Kids around the World. In Proceedings of FabLearn 2019; ACM: New York, NY, USA, 2019; pp. 104–111. [Google Scholar] [CrossRef]
  17. Dai, Y.; Liu, A.; Qin, J.; Guo, Y.; Jong, M.S.; Chai, C.; Lin, Z. Collaborative Construction of Artificial Intelligence Curriculum in Primary Schools. J. Eng. Educ. 2023, 112, 23–42. [Google Scholar] [CrossRef]
  18. Pranckutė, R. Web of Science (WoS) and Scopus: The Titans of Bibliographic Information in Today’s Academic World. Publications 2021, 9, 12. [Google Scholar] [CrossRef]
  19. Vieira, E.S.; Gomes, J.A.N.F. A Comparison of Scopus and Web of Science for a Typical University. Scientometrics 2009, 81, 587–600. [Google Scholar] [CrossRef]
  20. Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.; Brennan, S.E.; et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. BMJ 2021, 372, n71. [Google Scholar] [CrossRef] [PubMed]
  21. Bryman, A. Social Research Methods, 5th ed.; Oxford University Press: Oxford, UK, 2016. [Google Scholar]
  22. Choi, M. Design, Implementation, and Effects of Elementary Music Creation Class Using an AI-Based Music Program, Doodle Bach. Korean Music Educ. Soc. 2023, 52, 211–237. [Google Scholar] [CrossRef]
  23. Aktar Mispa, T.; Sojib, N. Educational Robot Kiddo Learns to Draw to Enhance Interactive Handwriting Scenario for Primary School Children. In Proceedings of the 2020 3rd International Conference on Intelligent Robotic and Control Engineering (IRCE), Oxford, UK, 10–12 August 2020; IEEE: Oxford, UK, 2020; pp. 87–91. [Google Scholar] [CrossRef]
  24. Li, X.; Xiang, H.; Zhou, X.; Jing, H. An Empirical Study on Designing STEM + AI Teaching to Cultivate Primary School Students’ Computational Thinking Perspective. In Proceedings of the 2023 8th International Conference on Distance Education and Learning, Beijing, China, 9–12 June 2023; ACM: Beijing China, 2023; pp. 1–7. [Google Scholar] [CrossRef]
  25. Joo, K.H.; Park, N.H. Design Artificial Intelligence Convergence Teaching and Learning Model CP3 and Evaluations. J. Curric. Teach. 2022, 11, 291. [Google Scholar] [CrossRef]
  26. Weiwei, S. Design of Auxiliary Teaching System for Preschool Education Specialty Courses Based on Artificial Intelligence. Math. Probl. Eng. 2022, 2022, 4504707. [Google Scholar] [CrossRef]
  27. Weng, T.-S.; Li, C.-K.; Hsu, M.-H. Development of Robotic Quiz Games for Self-Regulated Learning of Primary School Children. In Proceedings of the 2020 3rd Artificial Intelligence and Cloud Computing Conference, Kyoto, Japan, 18–20 December 2020; ACM: Kyoto, Japan, 2020; pp. 58–62. [Google Scholar] [CrossRef]
  28. Ma, Z.; Xin, C.; Zheng, H. Construction of a Teaching System Based on Big Data and Artificial Intelligence to Promote the Physical Health of Primary School Students. Math. Probl. Eng. 2021, 2021, 9777862. [Google Scholar] [CrossRef]
  29. Napierala, S.; Grey, J.; Brinda, T.; Gryl, I. What Type of Leaf Is It?—AI in Primary Social and Science Education. In Towards a Collaborative Society Through Creative Learning; Keane, T., Lewin, C., Brinda, T., Bottino, R., Eds.; IFIP Advances in Information and Communication Technology; Springer Nature: Cham, Switzerland, 2023; Volume 685, pp. 233–243. [Google Scholar] [CrossRef]
  30. Shi, Y.; Rao, L. Construction of STEAM Graded Teaching System Using Backpropagation Neural Network Model under Ability Orientation. Sci. Program. 2022, 2022, 7792943. [Google Scholar] [CrossRef]
  31. Shalileh, S.; Ignatov, D.; Lopukhina, A.; Dragoy, O. Identifying Dyslexia in School Pupils from Eye Movement and Demographic Data Using Artificial Intelligence. PLoS ONE 2023, 18, e0292047. [Google Scholar] [CrossRef] [PubMed]
  32. Chauca, R.; Simbaña, A.; Celi, C.J.; Montalvo, W. Virtual Learning Environment for Children with Down Syndrome Through Leap Motion and Artificial Intelligence. In Smart Technologies, Systems and Applications; Narváez, F.R., Urgilés, F., Bastos-Filho, T.F., Salgado-Guerrero, J.P., Eds.; Communications in Computer and Information Science; Springer Nature: Cham, Switzerland, 2023; Volume 1705, pp. 122–133. [Google Scholar] [CrossRef]
  33. Andinia, A.; Isnainiyah, I.N. Design of Learning Application Using Trivia Method Based on Google Assistant for Vision Impairment Disability. In Proceedings of the 2020 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), Jakarta, Indonesia, 19–20 November 2020; IEEE: Jakarta, Indonesia, 2020; pp. 220–225. [Google Scholar] [CrossRef]
  34. Huh, J.; Ann, S.; Hong, J.; Cui, M.; Park, J.Y.; Kim, Y.; Sim, B.; Lee, H.-K. Service Design of Artificial Intelligence Voice Agents as a Guideline for Assisting Independent Toilet Training of Preschool Children. Arch. Des. Res. 2022, 35, 81–93. [Google Scholar] [CrossRef]
  35. Huang, S. Design and Development of Educational Robot Teaching Resources Using Artificial Intelligence Technology. Int. J. Emerg. Technol. Learn. 2021, 16, 116. [Google Scholar] [CrossRef]
  36. Barnard, Y.F.; Sandberg, J.A.C. Applying Artificial Intelligence Insights in a CAI Program for “Open Sentence” Mathematical Problems in Primary Schools. Instr. Sci. 1988, 17, 263–276. [Google Scholar] [CrossRef]
  37. Villegas-Ch., W.; Jaramillo-Alcázar, A.; Mera-Navarrete, A. Assistance System for the Teaching of Natural Numbers to Preschool Children with the Use of Artificial Intelligence Algorithms. Future Internet 2022, 14, 266. [Google Scholar] [CrossRef]
  38. Podpečan, V. Can You Dance? A Study of Child–Robot Interaction and Emotional Response Using the NAO Robot. Multimodal Technol. Interact. 2023, 7, 85. [Google Scholar] [CrossRef]
  39. Williams, R.; Park, H.W.; Breazeal, C. A Is for Artificial Intelligence: The Impact of Artificial Intelligence Activities on Young Children’s Perceptions of Robots. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, Glasgow, UK, 4–9 May 2019; ACM: Glasgow, UK, 2019; pp. 1–11. [Google Scholar] [CrossRef]
  40. Wu, S.-Y.; Yang, K.-K. The Effectiveness of Teacher Support for Students’ Learning of Artificial Intelligence Popular Science Activities. Front. Psychol. 2022, 13, 868623. [Google Scholar] [CrossRef] [PubMed]
  41. Chen, S.-Y.; Lin, P.-H.; Chien, W.-C. Children’s Digital Art Ability Training System Based on AI-Assisted Learning: A Case Study of Drawing Color Perception. Front. Psychol. 2022, 13, 823078. [Google Scholar] [CrossRef]
  42. Shamir, G.; Levin, I. Neural Network Construction Practices in Elementary School. Künstl. Intell. 2021, 35, 181–189. [Google Scholar] [CrossRef]
  43. Wang, X.; Liu, Q.; Pang, H.; Tan, S.C.; Lei, J.; Wallace, M.P.; Li, L. What Matters in AI-Supported Learning: A Study of Human-AI Interactions in Language Learning Using Cluster Analysis and Epistemic Network Analysis. Comput. Educ. 2023, 194, 104703. [Google Scholar] [CrossRef]
  44. Wang, X.; Pang, H.; Wallace, M.P.; Wang, Q.; Chen, W. Learners’ Perceived AI Presences in AI-Supported Language Learning: A Study of AI as a Humanized Agent from Community of Inquiry. Comput. Assist. Lang. Learn. 2022, 37, 814–840. [Google Scholar] [CrossRef]
  45. Pareto, L. A Teachable Agent Game Engaging Primary School Children to Learn Arithmetic Concepts and Reasoning. Int. J. Artif. Intell. Educ. 2014, 24, 251–283. [Google Scholar] [CrossRef]
  46. Wu, C.-H.; Liu, C.-H.; Huang, Y.-M. The Exploration of Continuous Learning Intention in STEAM Education through Attitude, Motivation, and Cognitive Load. Int. J. STEM Educ. 2022, 9, 35. [Google Scholar] [CrossRef]
  47. Salas-Pilco, S.Z. The Impact of AI and Robotics on Physical, Social-emotional and Intellectual Learning Outcomes: An Integrated Analytical Framework. Br. J. Educ. Tech. 2020, 51, 1808–1825. [Google Scholar] [CrossRef]
  48. Toivonen, T.; Jormanainen, I.; Kahila, J.; Tedre, M.; Valtonen, T.; Vartiainen, H. Co-Designing Machine Learning Apps in K–12 With Primary School Children. In Proceedings of the 2020 IEEE 20th International Conference on Advanced Learning Technologies (ICALT), Tartu, Estonia, 6–9 July 2020; IEEE: Tartu, Estonia, 2020; pp. 308–310. [Google Scholar] [CrossRef]
  49. Khilmiyah, A.; Wiyono, G. Assessment of Emotional and Social Intelligence Using Artificial Intellegent. In HCI International 2023 Posters; Stephanidis, C., Antona, M., Ntoa, S., Salvendy, G., Eds.; Communications in Computer and Information Science; Springer Nature: Cham, Switzerland, 2023; Volume 1836, pp. 447–453. [Google Scholar] [CrossRef]
  50. Lee, M.R.; Yen, A.Y.-J.; Chang, L. Neurofeedback and AI for Analyzing Child Temperament and Attention Levels. In Knowledge Management and Acquisition for Intelligent Systems; Ohara, K., Bai, Q., Eds.; Lecture Notes in Computer Science; Springer International Publishing: Cham, Switzerland, 2019; Volume 11669, pp. 21–31. [Google Scholar] [CrossRef]
  51. Omokawa, R.; Matsuura, S. Development of Thought Using a Humanoid Robot in an Elementary School Classroom. In Universal Access in Human-Computer Interaction. Virtual, Augmented, and Intelligent Environments; Antona, M., Stephanidis, C., Eds.; Lecture Notes in Computer Science; Springer International Publishing: Cham, Switzerland, 2018; Volume 10908, pp. 541–552. [Google Scholar] [CrossRef]
  52. Lai, T.; Xie, C.; Ruan, M.; Wang, Z.; Lu, H.; Fu, S. Influence of Artificial Intelligence in Education on Adolescents’ Social Adaptability: The Mediatory Role of Social Support. PLoS ONE 2023, 18, e0283170. [Google Scholar] [CrossRef]
  53. Lin, X.-F.; Wang, Z.; Zhou, W.; Luo, G.; Hwang, G.-J.; Zhou, Y.; Wang, J.; Hu, Q.; Li, W.; Liang, Z.-M. Technological Support to Foster Students’ Artificial Intelligence Ethics: An Augmented Reality-Based Contextualized Dilemma Discussion Approach. Comput. Educ. 2023, 201, 104813. [Google Scholar] [CrossRef]
  54. Gupta, A.; Smith, A.; Vandenberg, J.; ElSayed, R.; Fox, K.; Minogue, J.; Hubbard Cheuoua, A.; Oliver, K.; Ringstaff, C.; Mott, B. Fostering Interdisciplinary Learning for Elementary Students through Developing Interactive Digital Stories. In Interactive Storytelling; Holloway-Attaway, L., Murray, J.T., Eds.; Lecture Notes in Computer Science; Springer Nature: Cham, Switzerland, 2023; Volume 14384, pp. 50–67. [Google Scholar] [CrossRef]
  55. Kajiwara, Y.; Matsuoka, A.; Shinbo, F. Machine Learning Role Playing Game: Instructional Design of AI Education for Age-Appropriate in K-12 and Beyond. Comput. Educ. Artif. Intell. 2023, 5, 100162. [Google Scholar] [CrossRef]
  56. Bingi, B.; Vardhan, H.; Singh, L. Interactive Learning System for Children. In Advances in Systems Engineering; Saran, V.H., Misra, R.K., Eds.; Lecture Notes in Mechanical Engineering; Springer: Singapore, 2021; pp. 419–426. [Google Scholar] [CrossRef]
  57. Rovai, A.P.; Wighting, M.J.; Baker, J.D.; Grooms, L.D. Development of an Instrument to Measure Perceived Cognitive, Affective, and Psychomotor Learning in Traditional and Virtual Classroom Higher Education Settings. Internet High. Educ. 2009, 12, 7–13. [Google Scholar] [CrossRef]
  58. Voulgari, I.; Stouraitis, E.; Camilleri, V.; Karpouzis, K. Artificial Intelligence and Machine Learning Education and Literacy: Teacher Training for Primary and Secondary Education Teachers. In Advances in Educational Technologies and Instructional Design; Xefteris, S., Ed.; IGI Global: Hershey, PA, USA, 2022; pp. 1–21. [Google Scholar] [CrossRef]
  59. Lozano, A.; Blanco Fontao, C. Is the Education System Prepared for the Irruption of Artificial Intelligence? A Study on the Perceptions of Students of Primary Education Degree from a Dual Perspective: Current Pupils and Future Teachers. Educ. Sci. 2023, 13, 733. [Google Scholar] [CrossRef]
  60. Su, J.; Zhong, Y. Artificial Intelligence (AI) in Early Childhood Education: Curriculum Design and Future Directions. Comput. Educ. Artif. Intell. 2022, 3, 100072. [Google Scholar] [CrossRef]
  61. Nunes Vilaza, G.; Doherty, K.; McCashin, D.; Coyle, D.; Bardram, J.; Barry, M. A Scoping Review of Ethics Across SIGCHI. In Proceedings of the Designing Interactive Systems Conference, Virtual Event. 13–17 June 2022; ACM: New York, NY, USA, 2022; pp. 137–154. [Google Scholar] [CrossRef]
  62. Wang, X.; Li, L.; Tan, S.C.; Yang, L.; Lei, J. Preparing for AI-Enhanced Education: Conceptualizing and Empirically Examining Teachers’ AI Readiness. Comput. Hum. Behav. 2023, 146, 107798. [Google Scholar] [CrossRef]
Figure 1. Process of selection of empirical studies.
Figure 1. Process of selection of empirical studies.
Education 14 00744 g001
Figure 2. Research objectives.
Figure 2. Research objectives.
Education 14 00744 g002
Figure 3. Learning content.
Figure 3. Learning content.
Education 14 00744 g003
Figure 4. Learning activity.
Figure 4. Learning activity.
Education 14 00744 g004
Figure 5. Learning outcomes.
Figure 5. Learning outcomes.
Education 14 00744 g005
Figure 6. Word cloud of learning outcomes.
Figure 6. Word cloud of learning outcomes.
Education 14 00744 g006
Figure 7. Pedagogy of activity or AI tool.
Figure 7. Pedagogy of activity or AI tool.
Education 14 00744 g007
Table 1. Inclusion and Exclusion criteria for SCOPUS-indexed articles.
Table 1. Inclusion and Exclusion criteria for SCOPUS-indexed articles.
Inclusion CriteriaExclusion Criteria
Empirical studies (peer-reviewed)Systematic reviews and meta-analyses
Studies focusing on implementations of artificial intelligence (AI) in preschool and primary school (students aged 4 to 12)Studies about AI literacy—teaching AI (except if teaching about AI occurred with the use of AI)
Sample: children in the age range of 4 to 12Sample: only teachers and/or parents
Scientific fields: computer science, social sciences, engineering, mathematics, psychology, physics and astronomy, decision sciences, arts and humanities, neuroscience, multidisciplinary, materials science, energy, Earth and planetary sciences, chemical engineering, environmental science, economics, econometrics and financeIrrelevant scientific fields: biochemistry, genetics and molecular biology, business, management and accounting, health professions, medicine
Written in EnglishAll other languages
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Aravantinos, S.; Lavidas, K.; Voulgari, I.; Papadakis, S.; Karalis, T.; Komis, V. Educational Approaches with AΙ in Primary School Settings: A Systematic Review of the Literature Available in Scopus. Educ. Sci. 2024, 14, 744. https://doi.org/10.3390/educsci14070744

AMA Style

Aravantinos S, Lavidas K, Voulgari I, Papadakis S, Karalis T, Komis V. Educational Approaches with AΙ in Primary School Settings: A Systematic Review of the Literature Available in Scopus. Education Sciences. 2024; 14(7):744. https://doi.org/10.3390/educsci14070744

Chicago/Turabian Style

Aravantinos, Spyridon, Konstantinos Lavidas, Iro Voulgari, Stamatios Papadakis, Thanassis Karalis, and Vassilis Komis. 2024. "Educational Approaches with AΙ in Primary School Settings: A Systematic Review of the Literature Available in Scopus" Education Sciences 14, no. 7: 744. https://doi.org/10.3390/educsci14070744

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop