Is AI Smarter than Chemistry Students?
Robyn Scott’s final year undergraduate project examined the relative capabilities of chemistry students and AI bots in solving set problems covering all areas of chemistry.
LLMs in general outperformed chemistry students, particularly so in questions based on either recall or numerical calculations. Students tended to excel at problems requiring visual interpretation, though the manner in which image-based data is presented to LLMs may be limiting AI’s performance . Our results suggest possibilities for redesigning both teaching and assessment. These should emphasise human strengths and ensure that LLM tools are used to enhance education rather than to short-circuit assessments.
Hi, this is a comment. To get started with moderating, editing, and deleting comments, please visit the Comments screen in the dashboard. Commenter avatars come from Gravatar.