Key points are not available for this paper at this time.
This research investigates the effectiveness of using questions generated by Chat Generative Pre-trained Transformer (ChatGPT), a large language model, to improve student comprehension and performance in number base conversions. A mathematics course with 340 participants underwent pre-assessment to establish the baseline knowledge. Following a lecture, a subgroup of 170 students received a formative assessment with varying difficulty levels of ChatGPT-generated questions. Post-assessment was administered to all students, aiming to discern improvements and evaluate the impact of ChatGPT-generated questions on learning outcomes. The Wilcoxon-singed rank test revealed a significant enhancement in student performance after the intervention, while the Man-Whitney U test showed that the students who were exposed to the ChatGPT-generated questions scored better than those who did not, and its effectiveness was not gender-biased. More importantly, the findings show that the group exposed to ChatGPT-generated questions not only performed better but also demonstrated a deeper understanding of the material compared to their peers. This suggests that ChatGPT's questions effectively targeted diverse learning needs, potentially by providing personalized learning experiences within the assessment. This research contributes to Artificial Intelligence (AI) integration in education, highlighting ChatGPT's potential to create effective and insightful learning experiences in mathematics courses. This research aims to equip students with essential skills and knowledge that are directly applicable to their academic and professional pursuits in computing and technology. The potential contributions of this research include enhancing students' computational proficiency, fostering innovative problem-solving abilities, and preparing them for the complex demands of the technology industry.
Prathigadapa et al. (Tue,) studied this question.