Integration of Artificial Intelligence and Machine Learning in Education: A Systematic Review

Review Article https://doi.org/10.12973/ijem.11.2.203 International Journal of Educational Methodology Volume 11, Issue 2, 203 - 216. ISSN: 2469-9632 http://www.ijem.com/ Integration of Artificial Intelligence and Machine Learning in Education: A Systematic Review Manuel Reina-Parrado* University of Sevilla, SPAIN Pedro Román-Graván University of Sevilla, SPAIN Carlos Hervás-Gómez University of Sevilla, SPAIN Received: January 30, 2025 ▪ Revised: March 12, 2025 ▪ Accepted: April 14, 2025 Abstract: This PRISMA-based systematic review analyzes how artificial intelligence (AI) and Machine Learning (ML) are integrated into educational institutions, examining the challenges and opportunities associated with their adoption. Through a structured selection process, 27 relevant studies published between 2019 and 2023 were analyzed. The results indicate that AI adoption in education remains uneven, with significant barriers such as limited teacher training, technological accessibility gaps, and ethical concerns. However, findings also highlight promising applications, including AI-driven adaptive learning systems, intelligent tutoring, and automated assessment tools that enhance personalized education. The geographical analysis reveals that most research on AI in education originates from North America, Europe, and East Asia, while developing regions remain underrepresented. Without strategic integration, the uneven implementation of AI in education may widen social inequalities, limiting access to innovative learning opportunities for disadvantaged populations. Consequently, this study underscores the urgent need for policies and teacher training programs to ensure equitable AI adoption in education, fostering an inclusive and technologically prepared learning environment. Keywords: Artificial intelligence, ChatGPT, education, machine learning, teacher training. To cite this article: Reina-Parrado, M., Román-Graván, P., & Hervás-Gómez, C. (2025). Integration of artificial intelligence and machine learning in education: A systematic review. International Journal of Educational Methodology, 11(2), 203-216. https://doi.org/10.12973/ijem.11.2.203 Introduction Increasingly, technologies are doing things that previously only humans could do. This is so until a time comes when I do practically all of them. This is what we have been calling technological globalization (Kuleto et al., 2021; RodríguezGarcía et al., 2020). In education, these technologies are having an amazing impact, enabling access to more and different educational resources (Hoosain et al., 2020). New online learning platforms and multimedia content are emerging to enhance teaching quality. These tools leverage artificial intelligence (AI)to analyze student performance, identifying patterns such as increased failure rates in specific tasks, prolonged response times in exams, or decreased engagement with the platform over time. By detecting these trends, educators can intervene more effectively to support student learning. Despite what it may seem, the incorporation of AI in the educational field is still in a developing phase, and is characterized by a slow adoption process. This is because emerging technologies tend to arrive in education after consolidating themselves in other sectors, such as production or social, and because there is a historical perception that teaching is a task that belongs only to human beings (Nicoletti & de Oliveira, 2020). Through different media, it has been possible to show that some professionals in the education sector are reluctant to incorporate AI (Chatterjee & Bhattacharjee, 2020; Kadhim & Hassan, 2020). Likewise, it is clear to think that AI represents a tool with enormous potential to address critical problems such as demotivation and school dropout (Salas-Rueda et al., 2020), challenges that significantly affect the current education system, especially since there are many teachers who are not able to provide solutions related to this issue, and AI can * Corresponding author: Manuel Reina-Parrado, University of Sevilla, Spain.  [email protected] © 2025 The author(s); licensee IJEM by RAHPSODE LTD, UK. Open Access - This article is distributed under the terms and conditions of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/). 204  REINA-PARRADO ET AL. / AI and ML in Education: Systematic Review provide them with points of view not contemplated until then. However, its application in this context remains an unexplored territory, offering multiple opportunities for innovation and improvement of educational processes. The field of AI in education is attracting increasing interest due to its innovative nature and the challenges faced by teachers in terms of their training in computational thinking. The lack of previous experience and the complexity of this discipline from its foundations make it crucial to explore how AI is being applied in educational contexts and what methods are most suitable to incorporate it effectively (Chang et al., 2022). To understand the current landscape, it is proposed to carry out a systematic review that analyzes the use of Machine Learning as part of AI. This approach will provide an innovative perspective on how these technologies are transforming the educational field, with the aim of preparing students to take advantage of the technological tools available in the future. According to Zawacki-Richter et al. (2019), the purpose of a systematic review is to answer specific questions using a structured, transparent, and reproducible search methodology, using clear inclusion and exclusion criteria to select relevant studies. This process includes coding and data extraction, which facilitates the synthesis of findings to identify both their practical applications and existing contradictions or limitations. The incorporation of advanced technologies such as AI in the classroom represents a complex and progressive process (Prendes-Espinosa & Cerdán-Cartagena, 2021). In this sense, a thorough review of the most recent research on the application of AI in education can offer a detailed and critical analysis of the current state of this emerging field. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement is developed as a guide intended to provide a standard approach to conducting systematic reviews. Its main purpose is to unify procedures, ensuring that the results obtained are consistent and useful for future research in the area of study (Page et al., 2021; Urrútia & Bonfill, 2010). Although PRISMA is not a systematic review in itself, it is an essential tool to carry it out in a rigorous and structured manner. PRISMA includes 27 elements that must be considered during the development of the research. These points allow for the generation of well-founded conclusions that reflect the state of knowledge on a specific topic, defined according to the selection criteria established for the review (Page et al., 2021; Urrútia & Bonfill, 2010). Since systematic reviews are dynamic, it is necessary to delimit a time frame that determines which articles will be included. However, it is recommended to update them periodically to incorporate new studies that expand and enrich the analysis (Page et al., 2021; Su et al., 2022; Talan, 2021; Urrútia & Bonfill, 2010). This research pursues the main objective of analyzing how AI and ML are integrated into educational institutions, examining the challenges and opportunities associated with their adoption. To achieve this, the study establishes the following specific objectives: to identify and compile key bibliographic sources related to the most outstanding publications in the field; and examine the findings of such publications to assess the impact of using artificial intelligence through ML-powered chatbots in education. In order to achieve these purposes, specific objectives have been defined that allow these issues to be addressed in a structured way through analysis: to explore the ways in which AI, through ML-based chatbots, is being implemented in the educational field; to investigate teachers' perceptions perceptions of the educational value of AI and students derived from the review on the integration of AI in the classroom; and identify the AI tools and programs most used in the educational context. Methodology This paper presents a systematic review of scientific publications focused on the use of AI in the educational field. For its development, the PRISMA methodology was used (Hutton et al., 2016; Page & Moher, 2017; Urrútia & Bonfill, 2010). PRISMA is structured into 27 elements that serve as a reference to ensure that systematic reviews are useful and understandable for readers (Hutton et al., 2016). The initial version of PRISMA, published in 2009, gained wide acceptance and application in various fields. However, the updated 2020 version, used in this study, introduces significant improvements, including the possibility of conducting dynamic systematic reviews, also known as "live", which can be continuously updated based on new data (Page et al., 2021). Database Selection and Article Selection The systematic review focused on three fundamental inclusion criteria: Machine Learning (ML), Education, and AI. The reason why these three criteria have been used was the following: a) ML is a fundamental branch of AI that allows machines to analyze data and learn from it to make predictions or make decisions. This criterion was included due to its growing impact on the development of educational tools and applications. ML techniques such as classification algorithms, regression, and neural networks are the basis of many systems for personalizing learning, adaptive assessment, and analyzing student behavior. Its inclusion allows us to International Journal of Educational Methodology  205 analyze how these technologies are being used in the design and implementation of innovative educational methodologies. b) The educational field is the key context of this review, as it seeks to explore how AI-based technologies are transforming teaching and learning methods. Including education as a criterion ensures that the selected studies are directly related to the impact of these technologies on educational institutions, pedagogical practices, and the training of students and teachers. In addition, this criterion helps to understand the specific benefits and challenges that educational communities face when incorporating AI into their processes. c) AI is the general framework under which applications such as ML and other subfields are developed. This criterion is fundamental because it allows us to identify research that not only deals with the practical use of AI, but also with its ethical, social and pedagogical implications in the educational field. By including AI as a criterion, a broader vision is guaranteed that encompasses both specific applications and theoretical reflections on its role in the transformation of education. Articles that met the three established criteria were selected for analysis. This selection process was carried out using databases internationally recognized for their relevance in the indexing of scientific literature, such as SCOPUS, Web of Science (WoS) and ERIC. The choice of these search sources is based on their relevance, coverage and international recognition in the indexing of scientific and academic literature. The combination of SCOPUS, WoS and ERIC ensures comprehensive coverage of relevant studies, balancing depth of analysis in the field of education (ERIC) with the breadth and quality of multidisciplinary publications (Scopus and WoS). This allows for a more complete view of how artificial intelligence and machine learning are impacting the field of education, while ensuring that the sources selected are rigorous and reliable. The search was carried out using a deductive approach, using keywords as the main filter and applying search strings based on Boolean operators, specifically: "Machine Learning" AND "Education" AND "Artificial Intelligence". The selected articles were exported to a spreadsheet in Excel format to facilitate their review and subsequent organization. Subsequently, they were transferred to an external platform for the management of bibliographic references: Mendeley (desktop version). This software, which is freely accessible, is designed to collect, organize and cite research. It allows data to be imported directly from compatible websites and recognized formats, which facilitates the management of bibliographic information (Barsky, 2010). Document Filtering and Selection Next, the results were limited to documents with access to the full text and published in final versions (excluding preprints, since they are not definitive and could be altered in the final publication). The inclusion/exclusion criteria were as follows: a) Inclusion criteria - Focused on ML as part of AI applied to the educational field. - Addresses the use of AI based on ML techniques. - Published between 2019 and 2023. - Applicable to any education system, without geographical or contextual restrictions. - Includes practical applications of AI or case studies that explore potential educational uses of these technologies. - It is limited to articles published in academic journals. - Written in Spanish or English. - Available in its entirety with full access to the text. - Final documents. b) Exclusion criteria: - It does not address machine learning or AI as main axes. - It is limited to dealing with a specific topic where AI is used only as a secondary tool to achieve other objectives. - Published in 2018 or in previous years. - Focused exclusively on a specific geographical context. - It does not include practical applications of AI or case studies that explore potential educational uses of this technology. 206  REINA-PARRADO ET AL. / AI and ML in Education: Systematic Review - It does not correspond to articles from academic journals. - Written in languages other than Spanish or English. - The article is not available for full reading. - Preprints. The selection of articles from 2019 onwards ensures that the included studies are representative of the most current technologies, methodologies and policies, maximising the relevance and impact of the results of this systematic review. It is precisely from 2019 that a notable increase in the adoption of AI-based tools in educational contexts has been observed. This period coincides with the rise of platforms such as ChatGPT, adaptive learning systems, and educational chatbots, making studies published in this time interval especially relevant for analysis. After this first filtering, the 297 results obtained are presented in Figure 1. Figure 1. Initial Screening The initial processing of the collected data was carried out using a spreadsheet in Excel format. For the SCOPUS and WoS databases, the procedure consisted of selecting the previously filtered articles and exporting them in CSV (Comma Separated Values) format, which is compatible with Excel and allows direct integration. In the case of ERIC, the export generates a file in nbib format, a file type used primarily in the PubMed database. This format is not directly compatible with Excel, so it was necessary to use the Zotero reference manager (Alonso-Arévalo, 2015). The PubMed database was not used for manuscript screening because its query could have incorporated studies with a bias towards biomedical applications of AI, which is not the main objective of the analysis. Once the results of the three databases were obtained in separate Excel format files, they were manually combined into a single document. This consolidation allowed the data to be unified into a single XLSX file, from which the subsequent review and analysis was carried out. Article Review The review began with a total of 297 articles (Figure 2), which were consolidated into a single Excel spreadsheet to facilitate their initial management. 0 52 000 52 0 111 6 18 14 149 34 53 9 0 0 96 0 20 40 60 80 100 120 140 160 Conference proceedings Journal articles Book chapters Comments Conference papers TOTAL ERIC SCOPUS WoS International Journal of Educational Methodology  207 Figure 2. Flow Diagram of the Phases According to the PRISMA Model The records were organized and those that were duplicate (44) were eliminated, noting the databases of origin for each article. After this process, 253 documents remained to continue with the analysis. The first filter applied consisted of selecting only articles published in academic journals, reducing the number to 164. Documents discarded at this stage were archived for possible future research related to this line of study. These 164 articles were then evaluated by reviewing their titles, abstracts, and keywords. We included or excluded them on the basis that they met the objectives of the review. After applying the inclusion and exclusion criteria, 112 studies were excluded due to unavailability of full text, irrelevance to the research objectives, or lack of empirical data. Following this process, a total of 52 articles were downloaded and managed using the bibliographic reference software Mendeley for detailed reading and evaluation, aligning with the principles of Open Science. During this comprehensive review, previously established inclusion and exclusion criteria were reapplied. Among the main reasons for discarding items were the following: - The articles dealt with AI and ML tangentially, focusing on the specific content that was sought to work with these technologies, which does not meet the criterion that the focus should be on AI and ML as central elements. - The studies could not be extrapolated to broad education systems, as they were limited to very specific contexts or conditions, failing to meet the criterion of being applicable to any education system. - Although they addressed topics related to the object of study, they did not include practical applications of AI or case studies that showed specific uses in the educational field, which contravenes the established criteria. Identification Screening Selection Identified records of: Database (n=3) Records (n=297) Records deleted before screening: Duplicate records (n=44) Excluded records (n=89) Reports requested for recovery (n=164) Records not recovered (n=112) Excluded reports: Through a thorough reading (n=27) Reports Evaluated for Eligibility (n=52) Revised Records (n=253) Studies included in the review (n=25) Identification of studies through databases and registries 208  REINA-PARRADO ET AL. / AI and ML in Education: Systematic Review - Some papers identified as case studies turned out to be systematic reviews, failing to meet the type of approach sought for this review. Finally, after this process, 25 manuscripts were identified that met all the inclusion criteria and were selected to be part of the analysis (Figure 2). Final Selection of Articles After applying the inclusion and exclusion criteria, and carrying out a detailed analysis of the selected manuscripts, 25 final documents were obtained. These were organized in a specific subfolder within the Mendeley bibliographic manager and, subsequently, exported to an Excel spreadsheet to facilitate their handling and subsequent analysis (Table 1). Table 1. First Screening No. Year Author Educational Level Title of the Article Database 1 2023 Billingsley et al. K-12 Can a robot be a scientist? Developing students' epistemic insight through a lesson exploring the role of human creativity in astronomy SCOPUS -- -- 2 2022 Jokhan et al. Higher Education Increased digital resource consumption in higher educational institutions and the artificial intelligence role in informing decisions related to student performance SCOPUS Wos -- 3 2022 Nuankaew Higher Education / General Self-regulated learning model in educational data mining -- -- ERIC 4 2022 Niyogisubizo et al. Not specified Title not available in references SCOPUS Wos ERIC 5 2022 Grunhut et al. Medical Education Needs, challenges, and applications of artificial intelligence in medical education curriculum SCOPUS -- -- 6 2022 Zammit et al. K-12 Learn to machine learn via games in the classroom SCOPUS -- -- 7 2022 Vir-Singh and Kant-Hiran Higher Education The impact of AI on teaching and learning in higher education technology SCOPUS -- -- 8 2021 Stadelmann et al. General / Hybrid The AI-Atlas: Didactics for teaching AI and machine learning on-site, online, and hybrid SCOPUS Wos -- 9 2021 Kuleto et al. Higher Education Exploring opportunities and challenges of artificial intelligence and machine learning in higher education institutions SCOPUS -- -- 10 2021 Lampos et al. Special Education / Autism An artificial intelligence approach for selecting effective teacher communication strategies in autism education SCOPUS -- -- 11 2021 Harati et al. General / K12 Assessment and learning in knowledge spaces (ALEKS) adaptive system impact on students' perception and selfregulated learning skills SCOPUS -- -- 12 2021 Action Not specified Title not available in references -- Wos -- 13 2021 Pu et al. General / Bibliometric Identification and analysis of core topics in educational artificial intelligence research: A bibliometric analysis -- Wos -- 14 2021 Kanglang Higher Education Artificial intelligence (AI) and translation teaching: A critical perspective on the transformation of education SCOPUS -- -- International Journal of Educational Methodology  209 Table 1. Continued No. Year Author Educational Level Title of the Article Database 15 2021 Druzhinina et al. Mathematics / General Development of an integrated complex of knowledge base and tools of expert systems for assessing knowledge of students in mathematics SCOPUS Wos ERIC 16 2020 Salas-Rueda et al. General / Higher Ed Impact of the web application for the educational process on the compound interest considering data science -- Wos -- 17 2020 Marques et al. K-12 Teaching machine learning in school: A systematic mapping of the state of the art SCOPUS -- -- 18 2020 Muniasamy and Alasiry Not specified Title not available in references -- -- ERIC 19 2020 RodríguezGarcía et al. K-12 / General LearningML: A tool to foster computational thinking skills through practical artificial intelligence projects -- Wos ERIC 20 2020 Kadhim and Hassan Higher Education Towards intelligent e-learning systems: A hybrid model for predicting the learning continuity in Iraqi higher education SCOPUS -- -- 21 2019 How & Hung K-12 / STEAM Educing AI-thinking in science, technology, engineering, arts, and mathematics (STEAM) education SCOPUS Wos -- 22 2019 RuipérezValiente et al. Higher Ed / MOOCs Using machine learning to detect 'multiple-account' cheating and analyze the influence of student and problem features -- -- ERIC 23 2019 Palasundram et al. Higher Education / Chatbots Sequence to sequence model performance for education chatbot SCOPUS -- -- 24 2019 Sharma et al. Higher Ed / General Building pipelines for educational data using AI and multimodal analytics: A 'grey-box' approach -- Wos -- 25 2019 Luckin and Cukurova General Designing educational technologies in the age of AI: A learning sciences-driven approach SCOPUS -- -- Screening Update In a first phase, the systematic review considered the articles available in the databases up to February 2023. However, before concluding the first report in July 2023, a second search was conducted to include articles published between the two periods, which had not been initially evaluated. This additional search followed the same criteria and procedures previously established, although the time range was adjusted to include only documents published in 2023. After applying the inclusion and exclusion criteria, two new articles were identified (Table 2) that met the requirements and provided relevant conclusions to the study. Thus, the final review included a total of 27 articles. Table 2. New Items Added No. Year Author Educational Level Title of the Article Database 26 2023 Gilson et al. Medical Education How does ChatGPT perform on the United States medical licensing examination? The implications of large language models for medical education and knowledge assessment SCOPUS -- -- 27 2023 Chung et al. General / AI Applications Technology acceptance prediction of roboadvisors by machine learning SCOPUS -- -- Figure 2 presents a cluster map generated with VOSviewer from the keywords extracted from the analyzed articles. This map shows the close connection between machine learning (ML) and artificial intelligence (AI), highlighting how 210  REINA-PARRADO ET AL. / AI and ML in Education: Systematic Review both concepts are interrelated and complement each other in the processing and classification of data using these technologies. In addition, Figures 2 and 3 shows that AI is linked to various subject areas, while ML is directly associated with the data that AI collects and processes. Connected to smaller nodes, such as "adaptive education" or "data processing," these terms can be inferred to represent specific applications or areas of interest related to ML and AI. Another identified cluster is composed of terms such as "STEM," "educational assessment," and "technology in the classroom," indicating that several articles specifically explore the application of AI and ML technologies in teaching and evaluation processes within educational contexts (How & Hung, 2019; Sharma et al., 2019). Another group of key words such as "adaptive learning," "personalized education," and "student engagement" reflects the growing research interest in AI-driven systems designed to tailor educational experiences to individual learner needs (Grunhut et al., 2022; Zammit et al., 2022). Finally, another group focuses on "ethical concerns," "teacher training," and "technological barriers," highlighting the challenges that educators face when integrating these technologies into their practices (Kadhim & Hassan, 2020; Singh & Hiran, 2022). Together, these clusters illustrate the diversity of research topics within the field and reinforce the multidimensional impact of AI and ML on education. Figure 2. Main Keywords Extracted from the Reviewed Studies on AI and ML in Education This reinforces the idea that ML and AI are interdependent components, underlining the relevance of this study and its contribution to the understanding of these technologies in the educational field. Figure 3. Map of the Relationship Between Articles. Made With VOSviewer Results After screening scientific manuscripts, a total of 27 studies published between 2019 and 2023 in various databases were analyzed. The results show that the main sources of information used were SCOPUS, Web of Science (WoS) and ERIC. The total number of studies per database has been: International Journal of Educational Methodology  211 - SCOPUS: 19 studies. - Web of Science (WoS): 10 studies. - ERIC: 6 studies. The temporal distribution reflects a progressive growth in the publication of research related to AI and ML: - 2023: 3 studies (Billingsley et al., 2023; Chung et al., 2023; Gilson et al., 2023). - 2022: 6 studies (Grunhut et al., 2022; Jokhan et al., 2022; Niyogisubizo et al., 2022; Nuankaew, 2022; Singh & Hiran, 2022; Zammit et al., 2022). - 2021: 8 studies (Druzhinina et al., 2021; Harati et al., 2021; Kanglang & Afzaal, 2021; Kuleto et al., 2021; Lampos et al., 2021; Pu et al., 2021; Stadelmann et al., 2021; Talan, 2021). - 2020: 6 studies (Kadhim & Hassan, 2020; Marques et al., 2020; Muniasamy & Alasiry, 2020; Rodríguez-García et al., 2020; Salas-Rueda et al., 2020). - 2019: 4 studies (How & Hung, 2019; Palasundram et al., 2019; Ruipérez-Valiente et al., 2019; Sharma et al., 2019). Following the analysis of the selected studies, a thematic classification was developed to align the results with the research objectives and to better understand how AI and ML are being integrated into educational institutions. Three main themes emerged, reflecting the diverse applications and challenges identified in the literature. This categorization also highlights the potential and limitations of AI and ML in educational contexts. 1) Curriculum development for AI education (7 studies): These studies focus on integrating AI literacy and computational thinking into educational curricula, particularly at the K-12 level. How and Hung (2019) proposed incorporating AI-thinking into STEM education, aiming to foster analytical skills from early stages. Marques et al. (2020) conducted a systematic mapping of machine learning teaching in schools, identifying a growing interest in practical AI education. Zammit et al. (2022) examined game-based learning approaches to teach AI concepts, demonstrating positive effects on student engagement. Similarly, Rodríguez-García et al. (2020) introduced the LearningML tool to promote computational thinking skills through AI projects, while Stadelmann et al. (2021) explored didactic strategies for teaching AI in hybrid environments. Talan (2021) reinforced the importance of including AI in education through a bibliometric study, and Kanglang and Afzaal (2021) critically examined the role of AI in translation teaching, stressing curriculum adaptation needs. 2) Implementation of AI and ML tools in educational platforms (11 studies): This theme includes studies that analyze the use of AI-powered tools and platforms designed to enhance learning experiences and educational processes. Palasundram et al. (2019) tested the effectiveness of chatbots in supporting student learning. Vázquez-Cano et al. (2021) developed a chatbot to improve Spanish punctuation skills, enhancing flexible learning environments. Kadhim and Hassan (2020) proposed a hybrid AI model to predict learning continuity in higher education. Salas-Rueda et al. (2020) demonstrated the impact of web applications using data science for teaching compound interest. Jokhan et al. (2022) analyzed AI’s role in digital resource consumption and decisionmaking regarding student performance. Harati et al. (2021) evaluated the adaptive ALEKS system’s impact on selfregulated learning. Additionally, Grunhut et al. (2022) and Gilson et al. (2023) studied AI applications in medical education, particularly the potential of large language models like ChatGPT in knowledge assessment. Nuankaew (2022) developed a self-regulated learning model based on educational data mining. Sharma et al. (2019) proposed AI and multimodal analytics pipelines, while Ruipérez-Valiente et al. (2019) applied ML to detect cheating behaviors in MOOCs. 3) Barriers and challenges to AI adoption in education (9 studies): The final group of studies focused on identifying obstacles to effective AI integration in education. Singh and Hiran (2022) emphasized the digital divide and lack of educator readiness as significant barriers. Druzhinina et al. (2021) explored the complexity of expert AI systems in mathematics learning environments. Pu et al. (2021) conducted a bibliometric analysis revealing geographical and educational level disparities in AI research coverage. Kuleto et al. (2021) examined challenges related to AI and ML implementation in higher education institutions. Biurrun (2023) highlighted societal-level concerns, such as national restrictions on tools like ChatGPT. Similarly, Murphy-Kelly (2023) discussed ethical risks and global calls for caution in AI development. Nicoletti and de Oliveira (2020) proposed MLbased models for dropout prediction, underlining the need for further research on equity and accessibility. Lampos et al. (2021) analyzed AI’s potential to support teachers in autism education, identifying the need for better integration strategies. Finally, Cruz-Jesus et al. (2020) addressed the use of AI to assess academic achievement, stressing the importance of considering contextual barriers.