Human and AI-generated feedback in higher education: A systematic review of effectiveness and student perceptions

This study aims to compare the feedback provided by human professors and ChatGPT on university students’ work and to report on students’ perceptions of both types of feedback. A systematic review was conducted following PRISMA 2020 guidelines. Databases research included Web of Science, Scopus, EBSCO, ACM Digital Library, and IEEE Xplore, with additional gray literature sources, until October 2024. Inclusion criteria were cross-sectional studies evaluating university students’ work, comparing feedback from ChatGPT with human professors. Data extraction was performed using a standardized form, and risk of bias was assessed with the Joanna Briggs Institute critical appraisal tool. A narrative synthesis of the results was made. PROSPERO registration number: CRD42024566691. This review included 8 studies with 461 students. ChatGPT feedback was detailed and rapid, while human feedback was valued for its personalization and emotional support. Students appreciated the detailed and immediate nature of ChatGPT feedback but noted its lack of emotional nuance and context-specific guidance. Human feedback was preferred for addressing individual learning needs and providing affective support. A combination of both types of feedback to maximize benefits. ChatGPT can assist human teachers by providing detailed and timely feedback to university students. However, human supervision is essential to ensure feedback is nuanced and contextually appropriate. A hybrid approach can optimize the learning experience in higher education. Further research is necessary to explore AI applications in educational settings and understand their impact on learning outcomes.