AI as a Study Partner in Mathematics Education: Evidence from a Multi-Program University Survey

The use of AI-powered math platforms in mathematics learning

Artificial Intelligence-powered technologies are prevalent today, especially in education. Numerous studies have shown an increasing number of students utilizing AI-powered math platforms (AI-PMPs) as supplementary tools for mathematics learning. However, there is limited information on how these platforms specifically benefit pre-service mathematics education teachers at a public higher education institution. To address the population, knowledge, and theory gap, this study aims to evaluate AI-PMPs and students' experiences using them. By employing a convergent parallel mixed-methods approach, the researchers will explore the relationship between the frequency of AI-PMP usage and the students’ mathematics performance using a correlational quantitative design, and assess the students’ sentiments about AI-PMPs through a descriptive qualitative design. Proportional stratified random sampling was used to select 92 respondents for the quantitative phase, while purposive sampling identified 19 participants for the qualitative phase. Data was gathered through survey questionnaires and one-on-one interviews. The findings revealed that the frequency of AI-PMP usage did not significantly affect students’ mathematics performance, highlighting various purposes, impacts, and challenges associated with these platforms. The study found that while AI-PMPs have an inconsistent effect on mathematics performance, they can serve as useful supplementary tools for enhancing learning. Therefore, it is recommended that students not overly rely on and put limits on using AI-PMPs. The study output was a user guide for the commonly used AI-PMPs.

ORIGINAL RESEARCH ARTICLE Generative AI in higher education: efficiency, motivation and challenges in teaching business mathematics

Generative AI in higher education: efficiency, motivation and challenges in teaching business mathematics

This study explores the impact of applying ChatGPT-4, a generative artificial intelligence (AI) model, in the teaching of Business Mathematics. The research had a dual aim: to generate diverse problem sets to support teachers, and to enhance students’ critical thinking by involving them in the verification of AI-generated solutions. Conducted with 342 undergraduate students at the University of Debrecen, the experiment focused on differentiation tasks. ChatGPT-4 solved these with 94% accuracy, and the majority of students (90%) also performed well. Student feedback indicated that the approach was both useful and motivating. Cluster analysis identified three distinct learner groups – Self-Determined Enthusiasts, Duty-Bound, and Drifters – who differed significantly in their engagement with AI-based learning. While most students positively evaluated the use of ChatGPT-4, many also recognised its limitations and the need for critical reflection. The findings suggest that the conscious and pedagogically grounded integration of AI into mathematics education holds considerable potential. However, the development of critical awareness and the continued presence of human oversight remain essential to ensure meaningful learning outcomes. However, its effectiveness depends on ethical use, ongoing critical reflection, and the sustained pedagogical involvement of educators.

Assessing Student Needs for AI-based Adaptive Learning in Higher Education

The study contributes user-derived design requirements that can guide the development of trustworthy and context-appropriate AI-supported learning platforms for undergraduate ICT students in programming-related courses at the two participating universities; broader generalization to other higher education fields requires further research.

Artificial Intelligence in Higher Education: A Transdisciplinary Analysis of Students’ Course Understanding and Perceived Academic Performance

This study investigates the relationship between students’ use of Artificial Intelligence (AI) in higher education, their understanding of the course content and their perceived academic performance. Using the transdisciplinary framework of Basarab Nicolescu and the Informing Science theory, we analysed survey data collected from 200 university students. A Chi-Square Test of Independence ($\chi^2(1, N=200) = 50.51, p < .001, \phi = .50$) shows that students who use AI for cognitive reasons, such as clarifying complex concepts, are almost three times more likely to report a deep understanding of their coursework than students who use it mechanically to generate text fast. Crucially, the data reveal a cognitive paradox: this increased understanding does not result in higher formal grades. These findings suggest that universities need to abandon a punitive compliance regime and reform traditional assessment models to better evaluate and support symbiotic human-AI learning.

Effects of AI guided experiential learning on market intelligence skills among university students in Colombia

This quasi-experimental study examined whether AI-guided experiential learning improves market intelligence capabilities among Colombian undergraduate students compared to traditional instruction. A total of 120 students from two universities participated in an eight-week intervention structured around Kolb’s experiential learning cycle. The experimental group used ChatGPT to support data interpretation, reflection, conceptual modeling, and analytical scenario testing, while the control group followed conventional teaching methods. ANCOVA results indicated significantly higher posttest scores for the AI-supported group, with a medium-to-large effect size (Cohen’s d = 0.64). Learning progression analyses showed the greatest gains during the reflective observation and abstract conceptualization phases, demonstrating that AI tools enhance analytical development when embedded within structured experiential processes. These findings suggest that AI integration can strengthen market intelligence education in resource-constrained environments by expanding access to advanced analytical capabilities while preserving human-centered pedagogical principles. Rather than advancing a new theory, the study offers context-specific empirical evidence—consistent with experiential learning and structured AI-scaffolding accounts—that AI-supported experiential activities can be implemented feasibly and are associated with meaningful short-term gains in market-intelligence performance under the conditions examined.