The release of OpenAI’s ChatGPT marked the beginning of the generative AI era, significantly impacting education. A study by researchers at the University of Pennsylvania explored how AI tools like ChatGPT affect math learning among nearly 1,000 high school students in Türkiye. Results revealed that while students using ChatGPT performed 127% better in practice exercises, their exam scores were 17% lower than peers without AI support. This discrepancy suggested that many students relied on AI for answers instead of genuinely engaging with the material. The custom “GPT Tutor” version, which provided hints rather than solutions, improved performance but still didn’t surpass traditional learning. These findings highlight potential downsides of generative AI in education, including reliance on incorrect information and diminished understanding. The study underscores the importance of integrating teacher guidance and structured AI use to enhance learning outcomes, emphasizing that generative AI may not yet be a viable educational solution.
Source link
