Answers are grounded in this paper and cite their sources.
Background: This study develops and evaluates ChemIA, a course-specific GPT that supports pharmacy students in undergraduate Medicinal Chemistry. Customised with instructor-authored materials, it better matches the syllabus and reduces generic-model inaccuracies. Methods: The exploratory evaluation assessed technical performance and student perceptions. Technical performance was tested with 15 multiple-choice questions from course assessments. Across repeated interactions, ChemIA largely matched instructor-validated answers; discrepancies mainly involved information found only in images. It also handled a small set of questions based on English-translated materials. These findings indicate basic multilingual functionality in the tested conditions, not comprehensive multilingual capability. ChemIA supported course activities including quizzes, pre-class study, podcast-style content, and a pharmacological case study. Results: Survey responses indicated that students perceived ChemIA as useful for clarifying medicinal chemistry concepts, preparing for examinations, and obtaining supplementary explanations. Students also reported greater confidence in ChemIA’s responses than in those generated by a generic GPT model. In the case-study activity, students rated ChemIA's explanations more often as very clear than those from the non-custom model. Conclusion: These findings offer exploratory evidence that a domain-grounded GPT can complement Medicinal Chemistry education. Its simple setup and adaptability may support similar applications in other disciplines. However, findings are limited by the exploratory design and the absence of direct measures of learning or academic performance.
No takes yet. Share an insight, caveat, or question.
Fookes et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: