Artificial Intelligence-Generated Feedback Versus Human Feedback in Medical Education: An Integrative Literature Review

Introduction: Feedback is central to medical education, and generative artificial intelligence (AI) has made individualized, timely, and scalable feedback possible. Questions remain, however, about its accuracy, contextual relevance, and capacity to promote learning compared with feedback from faculty and clinical supervisors. Objective: To analyze the evidence on AI-generated feedback compared with human feedback in medical education, considering its quality, educational effectiveness, applicability, and limitations. Methods: Integrative review structured with the PICo strategy. PubMed/MEDLINE, Embase, Scopus, Web of Science, and LILACS were searched for studies published between January 2020 and October 2026 in English, Portuguese, or Spanish. Studies were classified by level of evidence, and findings were synthesized through interpretive thematic analysis. Results: Eleven studies from Canada, the United States, Germany, Turkey, Saudi Arabia, and Japan were included: five randomized trials (level II) and six comparative descriptive or mixed-methods studies (level VI). AI-generated feedback generally matched or exceeded human feedback in comprehensiveness, consistency, and adherence to predefined criteria, especially in written, structured, and simulated tasks. Human feedback was superior in clarity, relevance, and accuracy in some studies and in more complex diagnostic and therapeutic situations. Learning outcomes were measured only in the short term, and hybrid models combining AI data with faculty input showed favorable results. Conclusion: AI can increase the availability and consistency of feedback but does not replace the clinical and pedagogical judgment of faculty. Supervised integration of both modalities is the most consistent approach, and studies with longer-term outcomes and Latin American settings are needed.