Transforming Student Revision: The Pedagogical Potential of Generative AI

The ability and pedagogical potential of ChatGPT, Google Gemini, and Claude AI to create study notes and practice questions were evaluated using undergraduate pharmaceutical science lecture content. A mixed methods approach examined quantitative metrics such as comprehensiveness scores as well as qualitative aspects including accuracy and clarity. Both basic and complex input prompts were used, and a measure of how these models processed different data formats in the form of visual lectures, statistical content, and descriptive teaching formats was also included. All demonstrated potential, yet their performance differed depending on the specific task and type of content. Claude AI produced study notes that were highly reliable, showed high comprehensiveness scores between 0.75 and 1.0. ChatGPT demonstrated exceptional ability to create practice questions at multiple cognitive levels whereas Google Gemini performed better with detailed prompts but faced difficulties with mathematical questions. The study demonstrated that detailed prompting led to higher output quality across all systems especially in generating practice questions at higher cognitive levels. Generative AIs prove to be beneficial educational resource creation tools when their specific attributes and constraints are considered. Their effectiveness depends on the task and subject matter, demonstrating that they work best when used as supplementary tools instead of standalone solutions. This research study advances knowledge about successful Generative AI integration in higher education and identifies areas needing additional development.