AI-Mediated Feedback in Computer-Aided Drafting Education: Exploratory Evidence on Student Performance and Perceptions

Large higher-education classes make timely feedback during practical sessions difficult, encouraging the use of artificial intelligence (AI) as a complementary resource in technical education. This study examined AI-mediated feedback during computer-aided drafting (CAD) activities using structured prompts and task segmentation, alongside instructor-provided feedback across alternating phases. Academic performance and student perceptions were analyzed in a classroom environment. A quasi-experimental within-subjects ABAB design was implemented with 34 first-year Industrial Engineering students over 20 weeks, as the curriculum progressed from two-dimensional (2D) drafting to three-dimensional (3D) modeling. Results showed significant performance variation, with lower scores in later phases involving greater procedural demands. Since feedback modalities were embedded within the curricular progression, these differences cannot be interpreted as isolated causal effects of feedback modality. Students rated both modalities positively and considered AI-mediated feedback useful for resolving specific doubts, although not more useful than instructor feedback, particularly in 3D tasks. AI-mediated feedback may complement CAD instruction, depending on task complexity, prompt precision, response reliability, and instructor guidance.