Generative artificial intelligence (AI) is increasingly embedded in higher education; however, its pedagogical role in assessment within medical education remains insufficiently characterized. This study analyzed medical students’ perceptions of using generative AI as a formative assessment strategy in a physiology course. First-year medical students enrolled in Physiology and Biophysics II at a Brazilian medical school participated during the 2023 and 2024 academic years (n = 156). In 2023, the activity was completed individually, whereas in 2024 it was conducted in small groups. Students used ChatGPT to generate concise, topic-focused texts based on self-constructed prompts and subsequently evaluated and, when appropriate, revised the AI-generated content. Perceptions were assessed using a structured questionnaire followed by guided feedback discussions. Students consistently described the activity as challenging, engaging, and educationally valuable. Prompt construction emerged as the most demanding component, underscoring the importance of aligning prompts with learning objectives. Participants also emphasized the need for critical appraisal of AI-generated outputs, identifying conceptual gaps and inaccuracies that required correction. Group-based activities were perceived as particularly beneficial for discussion and collaborative learning. Overall, the findings highlight the importance of explicit instructional guidance when integrating generative AI into assessment, emphasizing prompt design, critical reasoning, ethical use, and learner autonomy. When thoughtfully implemented, generative AI may serve as a catalyst for meaningful learning in physiology education rather than a shortcut to answers.
Montrezor et al. (Mon,) studied this question.