Algorithmic feedback vs. human feedback: A comparative study of generative AI in STEM assessment practices
Year: 2026
Type: article
Abstract: Generative AI has renewed interest in automated formative feedback, yet evidence from STEM assessment remains limited. This comparative study examined algorithmic and human feedback in undergraduate Biology, Chemistry, and Calculus courses. 126 students were assigned to a course to receive either large language model-generated feedback or instructor feedback on a draft constructed-response task. Outcomes included revision quality, transfer performance, expert audit ratings of feedback, and student perceptions. Algorithmic feedback was substantially faster and more extensive, but human feedback was more accurate, more actionable, more aligned with the rubric, and associated with greater revision gains and higher transfer scores. Students also reported greater trust, usefulness, fairness, and overall satisfaction with human feedback.
DOI: 10.32674/c2d27f86
Author: Joedel Peñaranda
Institution: Biliran Province State University
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W7164825586
