This systematic literature review examines research on artificial intelligence (AI) in mathematics education published between January 1, 2021, and August 1, 2025. Searches of Scopus and Google Scholar identified 922 records; after deduplication, screening, and full-text eligibility assessment, 42 peer-reviewed journal articles and conference proceedings were included. The review used descriptive quantitative summaries and a deductive-inductive thematic synthesis. Two independent reviewers conducted screening (Cohen's kappa = 0.88), and methodological quality was assessed using the Mixed Methods Appraisal Tool (MMAT). The included literature indicates increasing attention to generative AI, personalised support, feedback, teacher practice, academic integrity, and equity. Evidence for educational benefits varies substantially across study designs and contexts, and technical capability should not be equated with demonstrated classroom effectiveness. Geographic patterns in the selected sample are described without attributing them to regulatory, economic, or infrastructural causes that were not directly tested. Because the 2025 search covered only January through August 2025, publication counts are treated as partial-year data and are not directly comparable to complete prior years. Key limitations include reliance on two databases, English- and Russian-language restrictions, reproducibility constraints in Google Scholar, methodological heterogeneity, and limited long-term evidence. Overall, AI shows potential to support mathematics teaching and learning, but stronger longitudinal and comparative evidence is needed to establish effectiveness, equity, and sustainable implementation.
This article reports an exploratory quantitative field survey of teachers’ perceptions of STEM integration in Israeli technological education. The study analyzed 589 complete questionnaire responses from in-service teachers who participated in a programme-linked STEM and technological education context. The questionnaire examined pedagogical perceptions, engagement and motivation, innovative pedagogy, future readiness, implementation-related perceptions, and strategic support for STEM integration. After psychometric screening, these dimensions were treated as exploratory composites rather than validated latent constructs. The final measurement model showed acceptable to excellent internal consistency (Cronbach’s alpha = .75-.91), after Q16 was removed from the future-readiness composite and Q22 was reverse-coded according to its original negative wording. Descriptive findings indicated stronger endorsement of classroom-proximal dimensions, particularly engagement and innovative pedagogy, than of broad future-readiness claims. Group differences were observed across age, teaching experience, and academic qualification, but interpretation is bounded by the non-probability, programme-linked sampling frame. Welch ANOVA and Games-Howell comparisons were used where variance homogeneity was violated. Hierarchical regression with dummy-coded demographic controls showed strong within-questionnaire associations between strategic support for STEM and both innovative pedagogy and implementation-related perceptions. The negative adjusted coefficient for future readiness was interpreted as a suppression pattern under substantial overlap with implementation-related perceptions, not as a substantive negative relationship. The findings suggest that teacher support for STEM depends less on reform rhetoric alone and more on whether STEM is experienced as pedagogically usable, professionally credible, and organizationally feasible in school practice.
Visualization is central to mathematical cognition, though the study of visualization in mathematical education (VME) has taken diverse, interdisciplinary directions. This paper seeks to critically chart the intellectual landscape, conceptual development, and research nature of VME literature. A revised bibliometric search strategy was implemented in the Scopus database using an expanded TITLE-ABS-KEY query that combined visualization-related terms with mathematics-education-specific terms. After applying document type, language, publication-year, duplicate-removal, and relevance-screening criteria, a final corpus of 279 journal articles published between 2014 and 2024 was retained for analysis. VOSviewer and Biblioshiny were used to conduct bibliometric mapping, thematic analysis, and temporal evolution to analyze patterns of publications, journals, and authors that emerged and disappeared, as well as influential collaboration patterns, thematic clusters, and research directions. Results indicate that studies on VME have been gradually increasing, with major contributions from the social sciences, mathematics, and computer science. The findings also indicate uneven geographical distribution and varying thematic concentration across VME research. This research offers a synthesis of VME research, organized by structural, thematic, and temporal perspectives. Also, the findings contribute to future visualization-focused mathematics education research by supporting stronger theoretical integration, methodological soundness, and broader contextual and technological representation across diverse educational settings.
This study examines the impact of contextualized instruction supported by Python programming on students’ understanding of numerical sequences in upper secondary education. The research was conducted using a quasi-experimental pretest-posttest design with 48 informatics-profile students at IAAP “Andrea Durrsaku” in Kamenica, including an experimental group (n = 25) and a control group (n = 23), without random assignment. A mixed-methods approach was employed to combine quantitative results with qualitative insights into students’ engagement and learning processes. The experimental group engaged in real-world, Python-supported contextual tasks, while the control group followed traditional instruction. AI-assisted tools were introduced only during the post-test phase for the experimental group as supportive resources for code generation. Data were collected using pre-tests and post-tests, open-ended student questionnaires, and structured observation sheets. The data were analyzed using SPSS, and the results were subsequently described and interpreted. The results revealed a statistically significant effect of the instructional intervention on students’ post-test performance after controlling for pre-test differences. The experimental group significantly outperformed the control group and demonstrated a large effect size (F(1, 45) = 32.370, p < .001, Partial Eta Squared = .418). Students also demonstrated improved abilities in pattern recognition, constructing general terms, and interpreting results, along with increased engagement and participation in learning activities. Overall, the findings suggest that integrating contextualized learning with programming-based support can enhance students’ conceptual understanding of numerical sequences.
Engineering drawing is fundamental to technical and vocational education and training (TVET), yet many students struggle with orthographic projection as a foundational skill. Despite ongoing efforts in vocational education reform, limited attention has been given to how orthographic projection is taught and experienced in classroom practice. This study investigates the challenges students face, current teaching practices, and perceived opportunities for improving engineering drawing instruction in Indonesian vocational high schools. Using a qualitative design, data were collected through semi-structured interviews with ten students and five teachers. Reflexive thematic analysis revealed consistent patterns of student difficulties in spatial visualization, understanding projection methods, drawing accuracy, and the use of computer-aided design (CAD). Teaching practices were found to rely primarily on paper-based instruction and manual sketching, with limited integration of digital tools. Participants identified several potential strategies, including expanded CAD practice, the use of 3D-printed models, industry-linked projects, and immersive tools such as augmented and virtual reality. These strategies were described as helpful for supporting visualization and engagement in learning. These findings reflect participants’ perceptions of instructional needs rather than demonstrated effectiveness. The study contributes context-specific qualitative insights into learning bottlenecks and instructional practices in Indonesian TVET. It offers context-sensitive, tentative implications for instructional design and future research in vocational engineering drawing education.
The fast adoption of generative artificial intelligence (GAI) in higher education has led to the realization of the necessity to study the responses of educators as a professional group, although there is limited empirical research, especially in a new educational setting such as Azerbaijan. This pilot study is a quantitative investigation of the attitudes of Azerbaijani university teachers related to GAI, their adaptations in pedagogy, and the perceived difficulties and support requirements. The information was gathered through an online poll (n=30) in a university with a high level of research. Findings indicate that teachers are aware of the opportunities of GAI to personalize learning and administrative efficiency yet were rated moderately on AI literacy (Mean=3.42) and willingness to apply (Mean=3.21). Some of the key issues were academic integrity, the validity of assessment, and AI-assisted plagiarism. The exploratory analysis revealed that there was a good positive correlation between AI literacy and the perceived usefulness (r=0.759), where active adopters redesigned assessments and adopted process-oriented approaches. However, the conceptualization of institutional support was perceived to be inconsistent (Mean=3.04, SD=1.1). The results show that successful GAI implementation must involve contextualized professional growth and straightforward institutional policies that can resolve ethical and pedagogical issues. Although constrained by sample size, this research has given the first signs of the importance of educator-based support to facilitate responsible AI integration into the modernization of higher education.
In order to learn from and share good practices, this article analyses the experiences of Women’s University in Africa (WUA), and its success stories in reducing or removing structural barriers to women’s and other underrepresented groups’ participation in science. With funding from the International Development Research Centre (IDRC) we set to establish WUA’s good practices in gender mainstreaming in science. Data were gathered through a desk study of pertinent reports and University regulations, key informant interviews with important staff members, a gender audit questionnaire, and focus group discussions with students. The study found that the University has a policy of a female: male student ratio of 85%:15%. However, is not yet fully reflected in science, calling for interventions specifically targeted at increasing the uptake of science by female students. As a way of mainstreaming gender, the University adopted flexible study methods, such as face-to-face and open-distance learning among others.
Designing cognitively accessible mathematics instruction requires ensuring that learners can equitably process, understand, and apply complex mathematical ideas. This study examines students’ cognitive load when solving geometric integral problems by analyzing intrinsic, extraneous, and germane load within the framework of Cognitive Load Theory. Using a descriptive qualitative approach, data were collected from 144 university students in mathematics-related programs at a public university in Indonesia through think-aloud protocols, written solutions, and classroom observations. The data were analyzed using NVivo-assisted thematic coding and supported by radar visualizations to strengthen analytical trustworthiness. The findings indicate that intrinsic load constituted the dominant cognitive burden, primarily arising from symbolic-visual confusion and visual-spatial difficulties when students attempted to coordinate algebraic expressions with geometric representations. Extraneous load further increased cognitive demands due to ambiguous verbal instructions, misinterpretation of symbolic cues, and excessive technical language; however, visual scaffolding and self-generated sketches were found to reduce unnecessary processing. Germane load emerged among a smaller group of students who demonstrated conceptual transfer, reflective verification, and schema integration, indicating the development of mathematical reasoning as students connected geometric structures with their corresponding integral representations. Overall, the study highlights the central role of instructional design in regulating cognitive load to support meaningful conceptual understanding and reasoning in solving visually complex mathematical problems.
Design plays a central role in integrated STEM approaches. However, there is limited knowledge of how teachers from different S-T-E-M disciplinary backgrounds practice STEM design. To address this knowledge gap, we examined the STEM design practices of a convenience sample of 26 in-service teachers who voluntarily participated in a 7-month-long STEM professional development programme, divided into 4 Learning Community (LC) groups. As part of this programme, teachers designed STEM modules in the context of Nanoscience-Nanotechnology. Qualitative analysis of the LC discussions (synchronous), forum posts (asynchronous), and the designed artefacts provided insights both in terms of ideas/themes that teachers mostly discussed, as well as the design activity of the individual teachers. Frequencies of teachers’ inductively coded design actions during the LC meetings were noted, described through design visualisations and were used to infer the centrality of ideas discussed and the centrality of members’ activity. This analysis showcased that modelling, technicalities, robotics, and sensors were some common themes discussed among the four cases. Regarding the impact of teachers’ backgrounds on the nature of their practices, it was found that STEM design centrality was not restricted to any disciplinary background. However, the activity of most mathematics teachers appeared peripheral. Finally, teachers tended to contribute to parts of the artefact that were closer to their disciplinary expertise, while a few boundary-crossing design practices were noted.
STEM projects are integral to STEM education, fostering critical thinking, problem-solving, and creativity. However, students often struggle with topic selection, methodology, and iterative improvement. This study investigates the integration of Design Thinking (DT) into a year-long STEM project course in a Thai high school, examining how iterative loops impact project originality and feasibility. Using a qualitative approach, data were collected through proposals, progression reports, final reports, and presentations, all of which were analysed using originality and feasibility rubrics. Results revealed that students who engaged in iterative DT processes produced more innovative and practical solutions, while those with minimal iteration faced challenges in originality and feasibility. The findings highlight the importance of structured iteration and feedback in STEM education, providing insights for educators to optimize project-based learning and support effective students in STEM projects.
This qualitative study explores the support needed for the acquisition of digital research skills (DRS) in secondary STEM education. We approached this from both teachers’ and students’ perspectives, based on semi-structured focus group interviews. The findings reveal a strong consensus among participants on the importance of incorporating DRS into regular lessons. Teachers emphasize the need for better curriculum integration and specific support structures, while students primarily seek practical, step-by-step guidance for applying DRS. After combining these findings with a literature search, we propose ten suggestions to enhance support for guiding and developing DRS in secondary (STEM) education. These ten suggestions initiate concrete steps to bridge the gap between secondary and higher education in DRS. In doing so, we contribute to the ongoing discussion on strengthening essential skills in education and provide actionable insights for curriculum developers, educational institutions, publishers, tool developers, and teachers.
This study aims to explore AI trend studies in STEM education using documents from the Scopus Database. Between 2015 and 2025, 171 documents were selected by researchers using the PRISMA method. The R language, Biblioshiny, and VOSviewer were later used to conduct a comprehensive bibliometric study. This review identifies the dominant and emerging trends in artificial intelligence applications in STEM education. The findings reveal a significant increase in studies on the use of AI in STEM education, with the dominant themes being personalized learning, machine learning, and, more recently, generative AI, such as ChatGPT. The findings revealed that integrating AI into STEM education can significantly contribute to achieving SDG 4 (quality education). It encourages equity, accessibility, and educational excellence. At the same time, it supports SDG 8 (decent work and economic growth) by means of workforce development initiatives. The initiatives prepare students for high-quality jobs in AI economies where demand for AI skills has grown exponentially. Different countries, such as the USA, China, and Singapore, lead the research landscape but have unbalanced collaboration networks. SDG 10 (reduced inequalities) is a point of concern raised by the limited extent of research on a few nations. This is due to the possibility that economically and educationally diverse international settings might not be helped by AI-driven solutions and research agendas from rich contexts, potentially triggering increasing economic and educational inequalities. This research informs the strategic planning of AI-based applications in STEM education worldwide and provides insights into policymakers, scientists, and educators, while remaining aligned with the Sustainable Development Goals.
This study reports a preliminary classroom evaluation of a technology-supported instructional intervention for conservation learning in primary STEM education, using the culturally significant Bua Kheo Mongkol (Nymphaea khaomongkol; Thai: Bua Kheo Mongkol) as the local context. The pedagogical design and activity set were developed and refined through a multi-round Delphi process involving nine experts in educational technology, science/environmental education, and measurement/evaluation. The intervention was examined using a one-group pretest-posttest design with a two-week retention test involving 30 Grade 5 students. Knowledge outcomes were measured using the same 40-item multiple-choice test at pre-test (T1), post-test (T2), and retention (T3). Results showed a small but statistically significant increase in knowledge scores from T1 to T2, while T3 scores remained similar to T2, suggesting short-term maintenance of the modest gain. Post-intervention student feedback on usability, engagement, learning support, local relevance, intention to apply learning, and overall satisfaction was generally positive at the item level. Overall, the findings suggest that a culturally grounded, technology-supported classroom intervention was associated with modest knowledge improvement and positive descriptive student feedback. However, the one-group design and measurement limitations mean that stronger causal claims should be avoided, and future comparative studies are needed to test effectiveness more rigorously.
STEM education has become a cornerstone of 21st-century learning, sparking curiosity, creativity, and practical reasoning in young learners. In the early primary grades, STEM-based approaches help students bridge the gap between different disciplines and apply their knowledge to real-life challenges. This study aimed to design and implement a STEM-based instructional unit for grades one to three in Saudi elementary schools, examining its effectiveness in developing problem-solving skills across four key areas: identifying the problem, generating possible solutions, implementing those solutions, and evaluating the outcomes. Adopting a quasi-experimental design, the study involved 46 students randomly selected from three public elementary schools in Al-Ahsa, Saudi Arabia. The intervention consisted of a nine-lesson unit that integrated science, mathematics, engineering, and technology through activities tailored specifically to the developmental level of young learners. Data were collected using an oral problem-solving test aligned with STEM processes. The findings revealed that the STEM-based unit significantly boosted students’ overall problem-solving abilities. The most notable gains were seen in implementation skills, followed by problem identification, solution generation, and finally evaluation skills. These results suggest that hands-on, exploratory learning fosters a child's ability to apply and test their ideas even before they can systematically assess them. Ultimately, the study highlights how STEM-integrated learning cultivates inquiry, reasoning, and confidence in tackling real-world problems from the very start of schooling. Looking ahead, these results underscore the importance of expanding STEM instruction in Saudi primary education through teacher training and curriculum innovation
Attitudes towards science can influence students' academic performance. Therefore, the objectives of this study are to assess attitudes towards science and the impact on them of metacognitive strategies, self-efficacy, learning processes, trust in science, understanding of the nature of science, and gender. Quantitative ex post facto research was carried out with 147 Argentine first-year veterinary students. Four previously designed and validated questionnaires were used to assess each of the study variables. The scores obtained and the correlation, multiple regression, and mediation analyses suggest that: a) attitudes towards science were mediocre, and neither gender nor knowledge about the nature of science had a significant effect on them; b) monitoring, evaluation and planning of learning, and trust in science were the variables that most influenced the variability of these attitudes; and c) constructivist connectivity, science learning self-efficacy, learning risks awareness, and control of concentration had significant indirect effects on attitudes towards science.
The integration of STEM (Science, Technology, Engineering, and Mathematics) in physics education is gaining increasing global attention due to its role in developing 21st-century competencies such as critical thinking, creativity, and technological literacy. This study presents a systematic literature review (SLR) of STEM-based physics education research published in Scopus-indexed journals between 2016 and 2025. Following the PRISMA protocol, 57 articles meeting the criteria were analysed through thematic and descriptive synthesis. The results of the analysis indicate a significant increase in research activity after 2018, with a peak in publications occurring in 2023. Indonesia, Malaysia, and the United States emerged as the main contributors. This review also revealed that Project-Based Learning (PjBL) and Problem-Based Learning (PBL) are the most widely used pedagogical models, whilst technology-based strategies such as Arduino-based digital learning and the flipped classroom appear to be growing in prominence. Although STEM research is growing globally, there remains a gap in theoretical coherence, namely that most studies have not elucidated how the integration of STEM components (S, T, E, M) is applied in research. Furthermore, no studies have examined teachers’ digital readiness when implementing STEM in physics education, and there are limitations in studies across certain educational levels. Future research is recommended to use a longitudinal and mixed-methods approach to examine how STEM integration can enhance scientific literacy and higher-order thinking skills in physics education
The knowledge society faces challenges in addressing changes in students’ graduate profiles due to the rapid expansion of artificial intelligence, automation, and smart cities, which has generated the need for an educational paradigm centered on action and informed decision-making. Continuous teacher training constitutes the foundation of new educational proposals in the twenty-first century. This research was conducted through a systematic literature review following the PRISMA methodology applied to the field of education. A mixed-methods descriptive–analytical approach was adopted, supported by qualitative thematic and relational analysis aimed at identifying conceptual dimensions. A total of 96 articles were selected from the Scopus, ERIC, and Web of Science databases with the objective of analyzing the conceptual dimensions involved in the implementation of STEAM education in formal educational contexts. In conclusion, an integrated conceptual model derived from semantic network analysis inspired by Atlas.ti is consolidated. STEAM/STEM education is positioned as the structural core of the model, while teacher training is articulated as a mediating dimension that enables holistic pedagogical and didactic integration with teaching practice, in order to address challenges and leverage opportunities inherent to twenty-first-century education.
This quasi-experimental study investigated the effects of Guided Inquiry-Based Instruction (GIBI) assisted by Variation Theory (GIBI-VT) on the achievement and goal orientations of grade ten students in solid geometry. Participants (N=102) from Debre Tabor City, Ethiopia, were randomly assigned to: Experimental Group 1 (EG1, n = 31) receiving GIBI-VT, Experimental Group 2 (EG2, n = 39) receiving GIBI alone, and a Control Group (CG, n = 32) receiving the traditional teaching method (TTM). Quantitative data were collected using a validated mathematics achievement test and goal orientation questionnaire. The ANCOVA revealed statistically significant differences in mathematics achievement post-test scores. Additionally, the pairwise comparisons showed that both EG1 and EG2 significantly outperformed the CG. Moreover, MANOVA indicated EG1 scored significantly higher on mastery goals orientation than EG2 and CG. Finally, the regression analysis model revealed that student's gender, Performance-Approach Post-Test Scores (PApPostTS), GIBI-VT vs. TTM, and GIBI alone vs. TTM significantly explained 47.0% of the variance in their Mathematics Achievement Post-Test Scores (MAPostTS). The GIBI-VT is an effective strategy for enhancing students’ achievement and mastery goals orientation in solid geometry. The study recommends incorporating GIBI-VT into teacher training and curricula to address persistent challenges in geometry learning in Ethiopia and similar contexts
The assessment of STEM thinking in elementary education remains dominated by product-oriented measurement, often overlooking the lived processes through which students’ reasoning develops in classroom practice. This study employs a qualitative phenomenological approach to explore how elementary school teachers experience, interpret, and negotiate the assessment of STEM thinking within authentic learning contexts. Data were generated through in-depth semi-structured interviews with teachers experienced in implementing STEM-related activities and analyzed using phenomenological thematic procedures to reveal the essential structure of their experiences. Findings indicate that assessment is lived as (1) process-oriented yet constrained by product accountability, (2) centered on interpreting invisible, relational, and evolving student thinking, and (3) continuously negotiated between structural limitation and pedagogical possibility. These insights support a conceptual shift from viewing STEM assessment as outcome measurement toward understanding it as situated meaning-making shaped by teacher agency and classroom context. The study proposes a phenomenology of STEM assessment that contributes to global discussions on authentic evaluation, teacher professionalism, and 21st-century learning. Empirically, the findings highlight how teachers negotiate between institutional accountability and the evolving nature of students’ STEM thinking in authentic classroom contexts. Practically, the study underscores the importance of developing more process-sensitive and interpretive assessment practices in elementary STEM education
Lobachevsky geometry is an elective course that is difficult for students to learn. One of them is the concept of the number of angles in a triangle less than 180 degrees. The purpose of this study is to produce a design of a Triangle learning trajectory in Lobachevsky Geometry using the context of the Sumatran traditional snack "lupis cake". The subject of this study is a mathematics education student at one of the universities in Bengkulu province. The research approach used in this study is Design Research. This approach involves an iterative cycle consisting of three phases, namely the preparation phase, the experimental phase, and the retrospective analysis phase. The result of this study is that there are five activities in the learning trajectory of the Triangle in Lobachevsky Geometry using the context of Sumatran traditional snack "lupis cake". Using this context, students are able to find the concept of the number of angles on a triangle less than 180 degrees. The conclusion is that the design of the Triangle learning trajectory in Lobachevsky's Geometry in the context of the traditional Sumatran snack "lupis cake" is valid and practical for finding the number of angles in a triangle
