Essential Elements for Implementing AI Tools in Elementary School: A Systematic Literature Review

Abstract

The global use of Artificial Intelligence (AI) has attracted considerable attention, and its integration into educational systems is a priority that warrants further exploration. In collaboration with UNESCO, numerous organizations have proposed parameters advocating for the inclusion of AI in basic education systems. A systematic literature review (SLR) was conducted to identify these parameters from the existing research. Although these parameters have been mentioned in some studies, they are generally not prioritized in the research landscape. AI tools are primarily used to support students, while teachers typically employ a pedagogical approach centered on in-class activities. Additionally, essential conditions related to research requirements and involvement from the private and third sectors showed consistent adherence across the examined studies. However, it was found that only 52% of the studies included an ethical declaration regarding the data collected by AI during research development, especially regarding studies involving children. This review provides a guide for educational communities looking to enhance pedagogical practices through AI integration into their educational environments, but who may be uncertain about where to begin. Questions related to AI modality selection, pedagogical relevance, ethical considerations, and procedural guidelines for integrating AI into curricula are addressed through the insights provided in this review.

1. Introduction

Artificial Intelligence (AI) encompasses systems designed to simulate various aspects of human intelligence, including perception, learning, problem-solving, and manipulation, with a predominant emphasis on the convergence of machine learning and data analytics [1,2,3]. Given these characteristics, the future role of teachers in the context of AI integration has become a significant topic of discussion [4]. However, the integration of technological tools into educational settings is an area of growing interest and remains only partially explored.

In the 1970s, early forays into this field began with the development of a computer-assisted instruction system known as “Scholar” [5]. This system, which is based on a language model and text generation, is capable of generating questions and answers while leveraging a network of databases that includes facts, concepts, and procedures.

In primary education, one of the earliest studies was conducted by researchers from the University of Pittsburgh [6], who, over the course of a year, installed computer systems equipped with AI tools in primary schools. These systems functioned as assistants to teaching staff and institutional personnel, contributing to the development of computational, language, writing, and problem-solving skills among students.

Currently, there is an ongoing debate among researchers and educational authorities regarding the appropriate target audience for integrating AI tools into education. Kim et al. [7] argued that the discourse surrounding whether to prioritize AI in higher education or primary education has predominantly focused on higher education. Nearly a decade ago, it was recommended that AI literacy should begin in upper secondary education, where students can familiarize themselves with fundamental concepts and apply this knowledge to their prospective career paths [8]. However, despite the prevailing emphasis on higher education, there is growing recognition that introducing AI tools into primary education is not only feasible but also essential.

The rapid evolution of AI and its increasing societal impact suggest that delaying AI education until later stages can limit students’ foundational understanding and adaptability. As noted by Kim et al. [7], AI literacy should not be deferred because the advancing technological landscape necessitates early intervention to equip younger learners with the skills required to navigate future learning environments. Moreover, while classifications and taxonomies of AI in education, such as those proposed by Holmes [9], Tuomi [10], and Hintze [11], offer a structural understanding of AI applications, they do not address specific contextual challenges encountered in primary education. This gap underscores the critical need for empirical research focused on the effective implementation of AI tools in primary education.

Therefore, this study aims to fill this gap by conducting a systematic literature review (SLR) of the integration of AI tools in elementary schools, identifying pedagogical approaches, contextual conditions, and ethical considerations necessary for successful implementation.

This study focuses on primary education and its equivalents in other countries, such as K-6, which represents the six school years following the completion of preschool or kindergarten. This study examines a classification system for AI predicated on its use in educational contexts. Originally proposed in 2019 [9] and modified one year later by Tuomi [10], this classification system delineates AI into the following categories:

• Student instruction: this category focuses on Intelligent Tutoring Systems (ITSs), dialog-driven tutoring platforms, and language learning applications, which encompass features such as pronunciation detection.
• Student support: this category includes a variety of tools and systems that play essential roles, such as exploratory learning environments, formative writing evaluation mechanisms, learning network orchestrators, language learning applications, collaborative AI learning platforms, continuous AI assessment tools, AI learning companions, course recommendation engines, learning-by-teaching chatbots, and self-reflection support systems encompassing learning analytics and metacognitive dashboards.
• Teacher support: this category comprises a suite of tools and technologies, including ITS integrated with learning diagnostics, mechanisms for summative writing evaluation and essay scoring, platforms for monitoring student forums, AI-driven teaching assistants, systems for automatic test generation and scoring, content recommendation engines for open educational resources, plagiarism detection software, and tools for detecting student attention and emotion.
• System support: this category includes various tools and methodologies, such as educational data mining techniques for resource allocation, diagnostic frameworks for identifying dyslexia-related learning challenges, the utilization of synthetic teachers, and the application of AI in learning research endeavors.

Currently, there are several classification types for AI, such as the one proposed by Arend Hintze [11], which divides AI into two categories and two theoretical concepts that could be developed in the future: Reactive Machines, Limited Memory Machines, Theory of Mind, and Self-Awareness. Reactive Machines are basic AI systems that cannot form memories; these systems can only respond to specific situations occurring at a single moment and cannot access past or future information for decision-making; thus, they lack the capability to adapt to situations for which they were not designed [12].

Limited Memory Machines are AI systems capable of looking into the past and forming memories; thus, they can make decisions based on accumulated experience (Hintze, 2016). In other words, they require a database with simple and specific information to realize their functionality. This information is usually monitored, and their view of the past is only transient because it is not stored indefinitely [13].

Theory of Mind [11] refers to AI systems capable of understanding human thought and emotions, although they are currently unable to fully replicate this understanding. Currently, this type of AI exists only in a rudimentary form; however, advancements in the coming years may enhance its capabilities [14]. According to Hintze, the next evolutionary step in AI development is Self-Awareness, which remains confined to fiction. This concept refers to AI systems possessing the ability to be aware of their existence.

One of the most debated topics concerning AI is content generation. Various tools have provided an almost unlimited array of possibilities for the automatic production of texts, images, videos, sounds, music, and other media [15]. Generative AI is defined as a variety of tools that can create images, text, video, audio, code, science and other resources (Figure 1), often appearing as if they were generated by humans [16,17,18]. Notably, new generative AI tools are being released monthly, which makes it challenging to maintain an up-to-date classification of these technologies.

Although these classifications are intriguing and widely used, they do not align with the primary focus of this study. Regardless of the type of AI used, the pedagogical approach must be carefully considered. Based on UNESCO’s extensive experience documented across various countries [19], the pedagogical practices that have proven most effective in enhancing student outcomes are as follows:

• Educational classes or activities: these are instructional sessions led by teachers or facilitators, where knowledge is delivered verbally, through written materials, or through a combination of multimedia formats.
• Group work: involves collaborative learning strategies in which students work together to complete assigned tasks, encouraging engagement in solving complex problems and developing skills such as teamwork.
• Project-Based Learning (PBL): A student-centered educational method in which learners are guided by an instructor, and apply their skills to solve real-world problems over an extended period. PBL emphasizes student autonomy, goal-setting, teamwork, and research-based exploration of practical issues.
• Activity-Based Learning (ABL): ABL focuses on students progressing at their own pace through activities structured by educators. Typically conducted in a classroom environment, ABL fosters independence, exploration, and experimentation, culminating in project presentations. This approach involves active student engagement and collaboration.

In its publication [19], UNESCO outlined seven key conditions that are crucial for the successful implementation of elementary education programs:

• Need for Research or Analysis: refers to a distinct branch of educational discourse focusing on the investigation of requirements related to curriculum implementation.
• Development of Teacher Resources: this includes essential instructional materials such as textbooks and lesson plans, which are critical for effective teaching.
• Teacher Training: this encompasses the provision of specialized training tailored to the AI curriculum, along with the necessary resources to support this training, as highlighted in the relevant academic study.
• Hiring Staff or Capacity Building: this involves recruiting additional qualified educators to effectively implement the AI curriculum.
• Private Sector or Third-Sector Participation: This involves engaging external entities, often from the private or third sectors, to contribute to educational programs. In some regions, these entities may serve as part-time trainers or consultants.
• School Infrastructure Improvement: refers to the enhancement of hardware and internet access within schools to support the AI curriculum, including the installation of computer labs and servers necessary for the successful delivery of AI-focused education.
• Acquisition of Additional School or Classroom Resources: this refers to the procurement of classroom kits, coding tools, AI resources, and other materials designed to support teaching and learning activities.

Other perspectives on curricula that have successfully integrated AI into primary education highlight several important considerations. A comprehensive curriculum must include ethical guidelines for the safe and equitable implementation of AI in society, ensuring principles such as transparency, fairness, equity, nonmaleficence, accountability, and privacy [20]. Additionally, addressing the ethical and moral challenges of AI is essential because this can lead to the development of AI tools that promote ethical behavior [21].

A crucial aspect is AI literacy, which involves the foundational knowledge required to understand AI concepts, processes, and connections across various domains of AI [22]. It is equally important for educators and students to acquire the necessary skills to actively engage with AI tools, which will enable them to design and develop AI systems that address real-world challenges [23].

The ethical dimension is particularly significant. The integration of AI into educational systems requires collaborative efforts between governmental bodies, researchers, educational institutions, and broader societal stakeholders. However, before implementing educational initiatives, adherence to a robust set of ethical considerations governing AI usage is imperative. Several researchers [24] have explored the ethical considerations pertinent to teaching and learning with AI in education and three primary areas have been identified: data (including privacy, consent, and transparency), computational approaches (such as trust in algorithms), and the ethics of education itself (encompassing issues related to bias, teacher expectations, assessment validity, resource allocation, and pedagogical strategies).

The following outlines the ethical considerations relevant to leveraging AI to enhance teaching and learning [25]:

• Safeguarding the security and privacy of sensitive data collected by AI systems must be a top priority.
• Informed consent from both students and guardians is crucial, along with prior notification of data collection.
• Ongoing effects are required to mitigate algorithmic biases in AI systems.
• AI systems avoid discriminatory practices, regardless of origin, gender, ethnicity, or other characteristics.
• Transparency in the decision-making processes of AI systems is paramount.
• A clear chain of responsibility must be established for AI systems’ decisions.
• The human role in the classroom should not be replaced by AI systems.
• AI systems should meet accessibility requirements to ensure inclusivity.
• Educational programs should avoid overreliance on AI to prevent negative impacts on students’ critical thinking and emotional well-being.
• Human evaluators should assess students, with AI providing support when necessary.
• The ownership of data collected by AI systems should belong to students and their guardians.
• Educational curricula should include discussions on the ethical implications of AI use.
• AI systems should align with the objectives of educational programs and institutions.
• The development of educational policies is essential to ensure that AI systems adhere to ethical standards.
• Establishing ethics review committees is essential to evaluate and enforce compliance with ethical standards regarding AI in education.

Given the recent proliferation of AI tools (such as generative AI), their risks are still under exploration. This is particularly relevant in the context of basic education, making it crucial to analyze the ethical implications and risks associated with AI in K-6 educational research.

From the perspective of AI as an educational tool, Reiss [26] identified two major risks. First, there is the danger that teachers may become overly dependent on AI, neglecting to reflect on their educational objectives, which could result in less effective teaching. Second, there is a risk that AI-based educational systems might focus excessively on the acquisition of specific knowledge or skills, potentially sacrificing a more holistic educational experience.

An important consideration is the human factor, particularly the interaction between teachers and students, which includes emotional communication. While AI can effectively support theoretical content delivery, it may not provide sufficient emotional or motivational support, especially for students with low academic resilience [27]. Therefore, it is crucial for teachers to focus on fostering soft skills and competencies, such as communication, empathy, and emotional intelligence, which are essential for holistic student development. A compelling example highlighting the importance of human factors is the difficulty AI faces in supporting subjects like literature, as compared to more quantitative disciplines like physics [26]. This disparity arises because qualitative subjects often require interpretation and analysis within social and cultural contexts, necessitating an understanding of nuances, ironies, symbols, and contextual meanings that AI may not fully grasp. In contrast, quantitative subjects are governed by rigid structures and clear rules, making them more suitable for AI support.

From the above, it can be asserted that the integration of AI technologies into elementary schools has garnered international recognition as a progressive and forward-thinking subject. Accordingly, the current SLR addresses the following key questions:

• Which categories of AI tools are employed in elementary schools?
• What pedagogical approaches are utilized in the implementation of AI tools in elementary schools?
• What are the critical conditions for successfully implementing AI educational programs in elementary schools?
• What are the primary ethical considerations regarding the use of AI in elementary schools?

2. Method

To address the research questions in this investigation, we follow the methodology of an SLR using the PRISMA method, as well as a review of the method (see Supplementary Materials) proposed by Carrión and Serrano [28]. This approach involves the following steps: (1) Identification of the need for the review; (2) Formulation of research questions; (3) Selection of a bibliographic source; (4) Definition of keywords and search strings; (5) Establishment of inclusion and exclusion criteria; (6) Identification of relevant research; and (7) Data synthesis.

In line with the PRISMA method (Figure 2) [29], Scopus was chosen as the bibliographic source. The following procedure was used to select articles for synthesis. The search parameters used were TITLE-ABS-KEY (“AI,” “elementary school”) OR TITLE-ABS-KEY (“AI,” “K-6”), resulting in an initial yield of 169 articles (May 2024). The inclusion and exclusion criteria were as follows: Open access: All open access Language: English, reducing the pool to 52 articles. During this phase, the available filters on the Scopus platform were used. Additionally, a manual review was conducted to exclude duplicate articles and those that did not directly align with the research topic, leading to 25 valid articles for further analysis. This section of the process involved two researchers independently reviewing the articles to ensure consistency. The combination of automatic filters and manual review enhanced the heterogeneity of selected articles and minimized potential biases.

The robustness of this SLR is reflected in the findings derived from the research questions. Moreover, it highlights the limited number of studies on AI in elementary education, underscoring the need for more research at this level.

Scopus was selected as it is a reliable bibliometric data source, particularly for academic research in quantitative science. It offers high-quality data for research assessments, landscape studies, science policy evaluations, and university rankings [30]. In the education field, Scopus is recognized as a valuable resource with comprehensive coverage of educational and instructional technologies, given its comprehensive coverage [31].

The exclusion of over 50% of the preselected articles was largely due to misalignment between their content and the specific application of AI tools at elementary school (K-6) levels. Although many articles referenced both AI and elementary education in their introductions and theoretical framework, these elements were frequently omitted from the methodology. Furthermore, several studies were purely theoretical without practical application, reducing their relevance to this review.

3. Results and Discussion

The synthesis of the reviewed articles was organized into predefined categories derived from the existing literature, including those suggested by Tuomi [10] for categorizing AI tools, UNESCO pedagogical frameworks [19], and ethical considerations outlined by Devi et al. [25]. The data for each category were calculated based on the number of studies that aligned with each category and are presented in figures for clarity (Figure 3, Figure 4 and Figure 5). However, due to the extensive nature of ethical considerations, this category is not accompanied by a corresponding figure. Each category was populated according to the uses and approaches reported in the reviewed studies. Percentages and counts represent the distribution of articles within these predefined categories. The 25 reviewed articles (n = 25) encompass a range of diverse topics, and while each employs some form of AI tool, their methods of implementation and objectives frequently diverge.

3.1. Which Categories of AI Tools Are Employed in Elementary Schools?

To address this question, Tuomi’s categories [10] were employed because they are considered the most suitable for categorizing AI tools in educational settings. The findings are illustrated in Figure 3, which shows the distribution as follows: Student instruction (0%), Student support (60%), Teacher support (20%), and System or institution support (8%).

Figure 3. Categories of AI tools employed in elementary school. This figure includes a category for research studies that did not utilize any AI tools, either due to their nature or because they were deemed unnecessary.

Figure 3. Categories of AI tools employed in elementary school. This figure includes a category for research studies that did not utilize any AI tools, either due to their nature or because they were deemed unnecessary.

Additionally, 12% of the studies did not directly implement AI tools. These include one study that evaluated teachers’ perceptions of AI [32], another that examined competencies developed by teachers who implemented AI in their classes [33], and a third that assessed students’ understanding of AI [34].

The most prevalent category, explored by 16 articles (60%), was AI tools as Student support. These tools are often designed to assist students in developing problem-solving skills through programing activities [35,36,37,38,39], as well as employing chatbots [40,41], robots [42,43], text-generating AI [44,45], and, in some cases, integrated smart glasses [46,47] or other peripherals with specific software [48,49].

The subsequent category, AI tools as Teacher support, is represented by five studies. In these tasks, AI is employed to: (1) identify students with intellectual deficiencies [50]; (2) create sketches and simple images [51]; (3) serve as a tool to prevent school violence and bullying [40]; (4) develop programing activities [52], and (5) teach basic AI concepts with the help of a cleaning robot [53].

AI used as a support tool for institutions was documented in two studies, both of which utilized data processing to analyze the needs of students and teachers to improve curriculum programs [54,55].

Interestingly, no studies have focused on AI for student instruction, which may stem from concerns regarding the potential of AI replacing teachers. A debate exists between those who view this outcome as unlikely [56,57] and others who view it as a possible future scenario [58,59]. Despite this ongoing debate, all 25 reviewed articles agree that AI in elementary schools offers various benefits for students, teachers, and educational institutions.

Nevertheless, the integration of AI into elementary schools carries inherent risks. When students use AI, issues of cheating [60] and the potential hindrance of critical thinking and autonomy development [61] may arise. For teachers, risks include overreliance on AI, neglecting their educational responsibilities [26], and potentially overlooking interpersonal relationships with students [27]. At the institutional level, concerns about the security and privacy of personal data for both teachers and students [62,63], as well as uncertainties regarding the inherent biases in AI algorithms [27,64], must be addressed.

3.2. What Pedagogical Approaches Are Utilized in the Implementation of AI Tools in Elementary Schools?

To address this question, we applied the pedagogical frameworks proposed by UNESCO [19]. The findings revealed that the studies employed the following approaches: classroom or educational activities (40%), group work (20%), ABL (24%), and PBL (16%). Notably, 28% of the studies utilized more than one pedagogical approach, with 12% implementing all of them. Furthermore, 28% of the studies did not specify their pedagogical approach, and 32% did not incorporate pedagogy by nature (Figure 4).

Figure 4. Pedagogical approaches employed by elementary school teachers utilizing AI tools. This figure includes a category for research studies that did not employ any pedagogical approach, either due to their nature or because it was deemed unnecessary.

Figure 4. Pedagogical approaches employed by elementary school teachers utilizing AI tools. This figure includes a category for research studies that did not employ any pedagogical approach, either due to their nature or because it was deemed unnecessary.

In four of the studies [35,42,49,53], all the proposed pedagogies were explored, possibly because of their comprehensive assessment of the implementation of AI tools in educational settings. These studies conducted interventions with teachers and students over several weeks to facilitate the incorporation of various pedagogical approaches. Notably, one research project dedicated an entire module of its intervention to addressing the risks associated with using facial recognition on social media and the ethical considerations surrounding generative AI [49].

Among the studies that employ multiple pedagogical approaches, two stand out [38,43]. The first study [43] employed a conversational robot named Musio, which allowed teachers to integrate its use in classes so that students could practice conversational skills in English as a foreign language. This robot provides language personalization, enabling some students to feel more comfortable practicing with it and enhancing their confidence in speaking English.

The second study [38] was conducted in two sessions on the same day. The first session focused on teaching students basic machine learning, AI, and simple block-based coding concepts. The second session involved hands-on activities to allow students to develop algorithmic thinking, problem-solving, and creativity skills through block-based programing. This study underscores the critical role of teachers in developing activities that integrate knowledge and AI tools.

Lastly, among the studies that did not employ pedagogical approaches, one stands out, in which the competencies developed by teachers who integrated AI into the educational curriculum were measured [33]. This case highlights the formal implementation of AI in the educational curriculum and identifies the skills and competencies necessary for AI education. Based on the Technological Pedagogical Content Knowledge Model, it was found that teachers perceive themselves as more competent in pedagogical knowledge (i.e., designing and developing an AI curriculum aligned with school subjects, organizing the classroom in both online and in-person settings, providing timely feedback, and promoting peer review), technological knowledge (i.e., seeking relevant learning tools or materials for AI, demonstrating AI-based learning tools and materials in class, and being open to AI and new technologies), and content knowledge (i.e., understanding basic concepts and components of AI, grasping the fundamentals of computer science and Science, Technology, Engineering, and Mathematics [STEM], and understanding copyright when using open educational resources).

Other pedagogical approaches worth considering include gamification to teach AI comprehension topics and principles, which can be achieved through PBL and practical exercises for problem-solving using AI [65]. Additionally, active participation, subject integration, and PBL can facilitate meaningful and practical learning of AI-related concepts, allowing students to engage in hands-on and experiential learning [66].

Digital storytelling, which combines pedagogy and technology, enables students to learn new concepts while integrating different narrative elements [67]. This pedagogical approach can be employed to develop AI literacy in elementary schools [68] by helping students understand AI concepts and skills (e.g., what AI is, supervised learning, neural networks, AI-driven tools), as well as the importance of ethics in AI and its social impacts through various learning activities.

3.3. What Are the Critical Conditions for Successfully Implementing AI Educational Programs in Elementary Schools?

UNESCO [19] has delineated seven essential conditions (Figure 5): research or needs analysis (100%), development of resources for teachers (64%), teacher training (60%), staff hiring or capacity increase (0%), private sector or third-sector participation (100%), school infrastructure improvement (0%), and acquisition of additional resources for schools or classrooms (44%). Since the documents under analysis did not explicitly address these characteristics, elucidations and considerations are necessary to conduct the analysis.

Figure 5. Essential conditions for implementing AI tools in elementary school educational programs.

Figure 5. Essential conditions for implementing AI tools in elementary school educational programs.

First, the UNESCO considerations were not explicitly cited in the articles. While some studies describe them and are identifiable, in others, the conditions are not explicitly articulated but can be inferred from the research designs and educational processes.

Second, two UNESCO considerations apply to each of the analyzed articles: research or needs analysis and participation from the private and/or third sectors. The first consideration involves using these studies to conduct a needs analysis for integrating AI curricula in the respective countries. The second consideration refers to the involvement of individuals from the private and/or third sectors’ organizations in conducting and participating in the research.

Improvement in school infrastructure was not addressed in any of the studies, as such endeavors typically require significant investments and are predominantly undertaken by governmental agencies. The sole successful case of official AI implementation in a school curriculum [33] has not explicitly addressed the challenges encountered during its implementation. However, it can be assumed that the challenges encountered were similar to those mentioned by UNESCO, alongside potentially other hurdles given the unique needs of each educational system.

Apart from infrastructure enhancement, the acquisition of additional resources [39,41,42,43,44,46,47,48,49,53,69] poses a significant challenge because of the economic investments required. Notably, the procurement of a cleaning robot to teach students AI was highlighted [53]. In this experimental research project, an AI curriculum was developed to augment students’ scientific understanding from the perspective of “acting humanly” to “acting rationally.” The curriculum is organized into three themes: intelligent agents, sensors and machine perception, and machine learning. It is designed to capture and sustain students’ attention, interest, and motivation in learning AI.

Research aimed at developing resources for teachers and conducting teacher training is sometimes conducted concurrently [34,35,37,38,39,42,43,46,47,49,52,53,54,69]. Notably, one study sought to enhance interest and support student learning in STEM fields through AI use [42]. This study employed various AI-enabled robots over nine weeks, covering introductory AI content, an introduction to the robots used in the program, the anthropomorphism of the robot in speech and facial expressions (appearance, morality, emotions, and cognition), robot physics (movement programming and characteristics), voice and music composition using AI, creating images with AI, and developing an acting program for the robot. While the objective of enhancing interest in STEM topics and supporting learning was achieved, the students were not exposed to the risks and ethical dilemmas of AI use. The ethical approach was primarily intended to encourage students to treat robots ethically.

3.4. What Are the Primary Ethical Considerations Regarding the Use of AI in Elementary Schools?

Alarmingly, 48% of the analyzed studies did not include ethical statements. The remaining studies followed various ethical considerations: 8% anonymized participants’ data [37,48], whereas 44% informed participants of the study details and obtained signed consent from both participants and their guardians [35,36,37,38,40,42,46,47,48,49,53]. Furthermore, 20% declared adherence to international and institutional ethical codes [38,47,48,49,50], and 12% addressed the topic of “ethical considerations in the use of AI” during their research [33,42,49].

Studies that have delved into the ethical considerations of AI have done so in varied ways, reflecting their diverse objectives and study subjects. For example, one study [33] explored the competencies developed by elementary school teachers, categorizing ethics into three main competencies: understanding the essence of AI ethics (e.g., transparency, explainability, accountability, responsibility, fairness, privacy, predictability), understanding copyright laws, refraining from violations when utilizing open educational resources in the classroom, and comprehending sensitive concepts related to the AI workforce.

Another study conducted with K-5 students [49] proposed that students should understand the positive and negative impacts of AI on society. This was completed through activities such as open debates on the ethical and social repercussions of AI, like how image alterations can foster unrealistic beauty standards or how image recognition can propagate misinformation.

In the last of these studies, children aged 5 to 11 participated [42]. After a nine-week interaction, the students exhibited varied sentiments toward the robots, primarily empathy. This scenario prompted discussions about the attitudes children could adopt toward robots/AI and the ethical education they should receive at an early age to responsibly interact with AI.

It is understandable that, due to the nature of the research itself, ethical considerations regarding AI may not always be integrated into the methodological development of studies. However, it is concerning when studies involving children as participants do not address the ethical use of obtained data or fail to declare the obtaining of consent from guardians. Nonetheless, this study does not aim to discredit any of the analyzed studies. Instead, it acknowledges that the research was likely conducted with due consideration of the ethical responsibilities of the researcher and their affiliated institutions.

Other studies addressing the ethical implications of AI use in education, though not specifically focused on primary education, raise concerns and guidelines akin to those established earlier. Teachers are willing to explore, understand, and engage with the ethics of designing and applying AI in educational settings [24]. However, their role in teaching and learning raises ethical concerns regarding data privacy, discrimination, and transparency. These issues require clear communication and informed consent from participants and guardians to address them properly [25].

4. Conclusions

This SLR presents a comprehensive analysis of the current state of AI tool applications in elementary schools. The findings highlight not only the categories of AI tools employed, their pedagogical frameworks, and the conditions required for effective integration but also the ethical considerations associated with their use. Each section provides a detailed synthesis of the reviewed studies, contributing to a deeper understanding of the field. A notable finding was the limited availability of articles on the implementation of AI tools specifically in elementary or K-6 schools. The UNESCO initiative encouraging member countries to incorporate educational programs or curricula using AI tools in basic education is still in its early stages. Among the 11 countries that have successfully integrated curricula with AI tools, only China, Kuwait, Portugal, Qatar, and the United Arab Emirates have done so at elementary education levels [19].

Regarding the prevalent types of AI tools used in educational settings, this review reveals a preference for tools categorized as Student support, which have gained the most attention from researchers, followed by tools that assist teachers. However, significant gaps remain, particularly concerning the direct use of AI for student teaching without teacher intervention, which is largely unexplored because of concerns about AI replacing educators. Recent studies have suggested an increasing inclination toward tools employing generative AI, such as conversational language models, image-generating AI, and text-generating AI. This trend is expected to continue.

A central issue that may arise when teachers opt to integrate AI tools into their classes or courses is to determine the most appropriate pedagogy for this purpose. Unfortunately, only 48% of the research articles considered in this review utilized at least one pedagogical approach when incorporating AI tools into their courses. Furthermore, it would be beneficial to explore alternative pedagogies that specifically cater to the needs of students, teachers, and educational institutions.

Among the essential conditions for implementing AI tools in elementary or K-6 education curricula, three elements continue to challenge public education institutions, even without the involvement of AI: infrastructure improvement, acquisition of additional resources, and hiring new personnel. This implies that despite the intent to implement new technological tools and the willingness of researchers to offer free training to teachers along with educational and pedagogical materials for use with students, such efforts cannot be sustained over time without the corresponding institutional resources. In simpler terms, introducing technological tools during a research project to an institution that cannot afford to acquire its own tools will not lead to the lasting integration of those tools once the project has concluded, regardless of the positive outcomes achieved.

The integration of AI into elementary school curricula can yield benefits if executed properly, but it can also be detrimental if misused, potentially affecting students, teachers, and the institution. Understanding and adhering to ethical standards is essential for effective integration and responsible use. Nevertheless, there is an urgent need for clear and precise legislation governing AI usage, delineating the responsibilities of stakeholders within the educational sphere. The ethical implications of AI in education, particularly in primary schools, are a critical concern. Nearly half of the reviewed studies did not address ethical considerations, highlighting a significant gap in research involving minors. This underscores the need for comprehensive ethical guidelines and legislation, particularly regarding privacy and data security, to protect the interests of students and educators.

This SLR acknowledges several limitations. The first pertains to selecting the database. Although Scopus is widely respected and reliable, it does not encompass the entirety of educational research. As a result, valuable studies may have been excluded from the analysis simply because they were not indexed in this database. Another limitation is the focus on elementary education or K-6 education, which excludes studies that may provide unique insights into the use of AI at other educational levels. These limitations may introduce bias because databases and research indexes establish publication criteria. Similarly, focusing on a single educational level may create a bias that does not represent the full scope of the educational system or the reality of education.

Future research should explore generative AI tools and their integration into teaching practices with an emphasis on the role of teachers in effectively using AI rather than allowing AI to replace human educators. Addressing these gaps through empirical studies and careful planning will help ensure that AI’s role in primary education evolves responsibly and effectively, maximizing its potential to transform the educational landscape while upholding ethical standards.