Dongjo Shin , Hyejin Park , Sheunghyun Yeo

DOI:10.63311/jksmed.2026.29.3.121 Vol.29(No.3) 121-129, 2026

Artificial intelligence (AI), particularly generative AI, is rapidly reshaping mathematics education by creating new possibilities for teaching, learning, assessment, and teacher education. This editorial examines eight studies in this special issue to illuminate current directions in AI-integrated mathematics education and to consider their implications for the teaching and learning of mathematics. Four themes emerged from the studies: (1) mathematics teachers’ use of AI for instructional design and practice, (2) perceptions and acceptance of AI in mathematics education, (3) AI-supported preservice teacher education, and (4) AI-supported development of students’ mathematical thinking. Across these themes, the studies share a human-centered perspective in which AI is positioned not as a replacement for teachers’ professional judgment or learners’ mathematical reasoning, but as a resource for supporting and extending mathematical thinking and meaning-making through purposeful engagement with AI. The studies further indicate that the educational value of AI depends on pedagogically informed decisions regarding task design, prompting, feedback, and learner-AI interaction. They also underscore the importance of critically evaluating AI-generated outputs and strengthening teachers’ professional expertise. Collectively, this special issue highlights the need to conceptualize AI as a pedagogical mediator and as a thought partner whose potential is realized through purposeful instructional design and human professional judgment.

artificial intelligence, mathematics education

Amanda Gantt Sawyer , Zareen Gul Aga , Marcus Wolfe

DOI:10.63311/jksmed.2026.29.3.131 Vol.29(No.3) 131-152, 2026

Generative Artificial Intelligence (genAI) tools are being utilized in schools today to develop mathematics resources for the classroom; however, they can produce bias and incorrect mathematical information (United States Department of Education, 2023). Sawyer et al.’s (2020) Critical Curation Theory (CCT) provides a framework for describing how teachers can thoughtfully select, curate, and adapt resources based on their pedagogical knowledge, content knowledge, personal experiences, and the lesson's purpose. However, CCT was created prior to the introduction of genAI; thus, we propose an extension of the CCT to account for teachers’ ability to interact with the genAI tool to curate resources, make adaptations, and support synthesis, called CCT-GenAI. This interaction with GenAI requires a new set of skills to communicate with the program, which we refer to as Mathematics Education Prompt Engineering (MEPE). By expanding on CCT, we provide the MEPE as actions that afford a window into teachers’ actions through CCT-GenAI. Using the CCT-GenAI framework can guide teachers in communicating with genAI efficiently, allowing it to understand the teachers’ expectations, and aims to enhance mathematics teachers' ability to communicate and interact with the tool to create a new document.

generative artificial intelligence, prompt engineering, mathematics education prompt engineering, critical curation theory

Cristian Nava-guzmán

DOI:10.63311/jksmed.2026.29.3.153 Vol.29(No.3) 153-174, 2026

This manuscript reports a qualitative self-study of the situated use of a conversational AI during the planning, conduct, and formative evaluation of upper-secondary mathematics lessons. From a corpus of 102 records gathered across nine sessions (April-August 2025), we applied conventional content analysis to code in-class decisions and observable effects. The study documents how the chat was used for four classroom needs: (a) to build content familiarity before teaching specific mathematical topics such as the historical evolution of real numbers and proportionality relationships; (b) to shape and reduce preparation and editing effort for tasks involving contextualized problem-solving; (c) to secure continuity across sessions through reusable prompts and summaries; and (d) to regulate the level of directiveness during public explanation, pausing or constraining prompts while institutionalizing ideas at the board. The main outcome is a testable rule of action: use the AI in planning to open sense-making trajectories and anticipate likely stumbling blocks; pause it during whole-class institutionalization to protect pacing, turn-taking, and mathematical authority; and resume it to sustain continuity and gather formative evidence. The paper concludes with an examination of the contributions and limitations of a single-case design, as well as practical considerations for applying this approach to teacher education and school routines, including record-keeping and peer review.

self-study, conversational AI, mathematics teaching, classroom decision-making, instructional directiveness, upper-secondary education

Maximiliano Izzi-prato , Julieta Tejería-russi , José Gerardo Piña-aguirre , Armando Agustín Chávez-salazar , Asuman Oktaç

DOI:10.63311/jksmed.2026.29.3.175 Vol.29(No.3) 175-200, 2026

Although recently there has been an increase in the number of studies on the use of Artificial Intelligence (AI) in various educational settings, there is a lack of research in mathematics education about the role that text-to-image AI tools can play in the teaching/learning process. In this article, epistemic and pragmatic capabilities of a text-to-image AI in the context of mathematics education are examined. In order to explore how teachers might make use of these characteristics, we held a workshop aimed at presenting how this technology works, in which 10 teachers from different educational levels in Mexico participated. The teachers were invited to design mathematical activities making use of the tool. Eight of them submitted their proposals, which were then analyzed to identify the pragmatic and epistemic values that the teachers employed to design the activities. Derived from a thematic analysis, we present four different categories that summarize different ways in which the pragmatic and epistemic values of a text-to-image AI tool can be appreciated in the designs of the participating teachers. Our study illustrates that, due to their specific pragmatic and epistemic values, text-to-image AI tools have the potential to foster environments in which teachers can conceive and construct mathematical activities in novel, creative, and meaningful ways.

epistemic value, pragmatic value, epistemology, text-to-image AI, task design, teacher education

Eleni Tsami

DOI:10.63311/jksmed.2026.29.3.201 Vol.29(No.3) 201-218, 2026

The integration of Artificial Intelligence (AI) into higher education has sparked increasing debate regarding its role in teaching and learning. This study investigates contemporary trends in education following the introduction of AI tools and assistants at the University of Piraeus, a pioneer in technological innovation within the academic sector. A survey was conducted during the spring semester of the academic year 2024-2025 among 230 undergraduate students of the Department of Statistics and Insurance Science, focusing on their perceptions of AI in the teaching of Probability Theory. The results reveal that while the majority of students (aged 18-24) actively use AI in their daily lives, they remain skeptical about its reliability and maturity for exclusive use in academic contexts. Most respondents expressed a preference for traditional instruction delivered by professors, citing the irreplaceable human role in explanation, guidance, and academic dialogue. Nevertheless, students did not dismiss the value of AI entirely; instead, they favored a hybrid teaching approach, where AI is employed as a complementary resource to support, but not replace, professors. Furthermore, the findings highlight a strong demand for structured education on AI applications in everyday life, scientific research, and professional practice. Overall, the study suggests that while students recognize the potential benefits of AI in education, they advocate for gradual, balanced integration. Universities are therefore encouraged to adopt hybrid pedagogical models that combine technological innovation with human-centered teaching, while also fostering AI literacy as a key competence for future graduates.

artificial intelligence, teaching probability theory, students, Hellenic University

Ruthmae Sears , Cynthia Castro-minnehan , Orly Buchbinder , Yvonne Franco , Stephanie Arthur , Sandra Vernon-jackson

DOI:10.63311/jksmed.2026.29.3.219 Vol.29(No.3) 219-246, 2026

This study explores the perspectives of eight US secondary mathematics in-service teachers enrolled in a graduate program on the use of GenAI for mathematics teaching and learning, and addresses a knowledge gap for the United States population. Data were gathered from a questionnaire and a written survey to capture reflections from the workshop on Integrating GenAI in secondary mathematics. The Likert-scale survey items were analyzed using descriptive statistics. Open-ended data were analyzed using a thematic analysis based on two frameworks: the AI Competency Framework aspects (UNESCO, 2024) and the Nine stages of student learning and how AI can be used to enhance each stage (NSLEAI) (Pagani et al., 2025). We also employed open coding to identify other emergent themes, such as the role of AI, comfort and confidence levels with GenAI, and insufficient knowledge to use GenAI. The results suggest that participants primarily focused on AI pedagogy, AI foundations and applications, ethics of AI, and AI for professional development, while rarely attending to a human-centered mindset. The benefits of GenAI generally appeared alongside pedagogical considerations, whereas concerns were mostly related to ethical implications. Furthermore, Pagani et al.’s (2025) NSLEAI stages of student learning related to AI were seldom addressed.

GenAI, in-service teachers, graduate students, perspectives, secondary mathematics

Rosa Isela González-polo , Apolo Castaneda

DOI:10.63311/jksmed.2026.29.3.247 Vol.29(No.3) 247-271, 2026

Teacher education faces the challenge of connecting mathematical learning to socially relevant issues such as nutrition and health care. This study explores how a gamified proposal that uses Generative artificial intelligence (Generative AI) connects algebraic operations to health-related decisions among preservice primary education teachers. The intervention was conducted in a one-hundred-minute session with eighteen preservice teachers from a public teacher education college in Mexico, using narrative-based missions inspired by Nutrition Detectives and supported by a Generative AI tutor that provided differentiated feedback. Directed qualitative analysis of the preservice teachers’ digital dialogues with the tutor, the notes each team recorded in its notebook, and classroom observations revealed the following four findings across the three research questions. The narrative served as a structural condition for algebraic reasoning; the analytical, structural, and functional dimensions of algebraic thinking were mobilized interdependently; errors functioned as a resource for exploration when mediation shifted the focus toward rereading the problem; and individualization with Generative AI requires precise instructions in the prompt to differentiate mediation according to preservice teachers’ levels of algebraic reasoning.

gamification, Generative AI, algebra, preservice teachers

Nirmala Naresh , Terrie Galanti , Zuhal Yilmaz

DOI:10.63311/jksmed.2026.29.3.273 Vol.29(No.3) 273-300, 2026

This design-based research (DBR) project examines one mathematics teacher educator’s (MTE) iterative integration of generative AI tools in secondary mathematics content courses. The study is part of a multi-year collaboration investigating how prospective mathematics teachers (PMTs) learn to use AI for mathematical problem solving. Our ongoing research into PMTs’ engagement with AI tools informed successive instructional refinements. We use DBR to study how an MTE’s integration of generative AI tools evolves across four semesters, and what these integration decisions reveal about MTE’s instructional reasoning. We analyzed instructor reflections, interview data, scholarly outputs, and course artifacts to understand the progression of MTE decisions about problem selection, structuring AI-learner interactions, and scaffolding reflections. Over four semesters, the MTE’s focus shifted from exploring available AI tools to supporting mathematical problem solving with AI and ultimately to preparing PMTs to leverage AI responsibly in future classrooms. This work shows how DBR enables MTEs to make evidence-informed adaptation to their use of AI tools in classroom practice. We encourage other MTEs to employ DBR methodologies to respond to rapidly changing AI capabilities and refine instructional decisions based on classroom evidence. Potential directions for research include (a) broadening mathematics educator perspectives about the role of AI in mathematics learning; (b) building a case library grounded in student work to make AI engagement an explicit object of instruction; and (c) strengthening opportunities for mathematical sense making with AI integration.

artificial intelligence, mathematics teacher educator, mathematics education, design-based research, learner-AI engagement

Cristian Nava-guzmán

DOI:10.63311/jksmed.2026.29.3.301 Vol.29(No.3) 301-325, 2026

This exploratory design study analyzes the semiotic mediation of a conversational generative artificial intelligence (AI) tutor in the development of a high school student’s covariational reasoning. Through the implementation of the “bottle problem,” the research examines how a prompt design grounded in an “anti-solution” clause and an adaptive dialogue structure transforms traditional educational dynamics. By triangulating student artifacts of covariational reasoning, chat logs, and semi-structured interviews, the study evidences a significant cognitive transition in the subject, who evolved from pre-coordination (Level 2) to chunky continuous covariation (Level 5). The results suggest that the AI’s deliberate withholding of direct answers generates productive cognitive friction, which, far from hindering learning, compels the student to decompose the dynamic phenomenon into geometric intervals and coordinate rates of change. Consequently, it is argued that this approach reconfigures the student’s epistemic agency, positioning the AI not as an oracle but as a reflective partner that scaffolds the construction of complex mathematical arguments.

covariational reasoning, AI conversational tutor, formative feedback, prompt architecture