JOURNAL OF EDUCATIONAL RESEARCH, 2025 (SSCI)
The aim of this study is to evaluate the capacity of ChatGPT-4 and Gemini chatbots in detecting and addressing students' misconceptions related to exponential expressions. Responses from nine students in southern T & uuml;rkiye to a conceptual test on exponential expressions were analyzed by these chatbots. The findings indicate that both ChatGPT-4 and Gemini were partially successful in identifying and correcting students' misconceptions. However, ChatGPT-4 outperformed Gemini in this area. Both chatbots were able to provide appropriate guidance in certain cases. Nevertheless, ChatGPT-4 demonstrated superior ability in interacting with students and responding to their inputs. On the other hand, Gemini was more effective in offering accurate solutions and feedback. The study reveals that chatbots may encounter difficulties in detecting and correcting conceptual misunderstandings. To ensure the effective use of chatbots in educational settings, the study recommends the integration of synchronous and asynchronous classroom environments in which teachers can closely monitor student responses.