Humanizing AI Feedback: Enhancing Self-Efficacy and Creative Agency in Design Education

Abstract

Artificial intelligence (AI) is increasingly used in higher education,  however, little is known about how AI feedback can be designed to  support the relational and affective dimensions of studio learning. This  mixed-methods pilot study examines an AI feedback system intentionally  crafted with empathic and partnership-oriented features to assist senior  design students in a studio course. Drawing on social-cognitive theory  and feedback literacy, we propose an AI-Empathic Feedback Design  model that positions AI as a relational co-participant in studio feedback  processes. Twenty final-year design majors used a prototype tool  providing qualitative critiques, sentiment-aware prompts, and progress  dashboards over a 15-week semester. Pre/post self-efficacy scores  increased from 3.40 to 4.10 (Cohen’s d ≈ 1.20), and students engaged in  more iterations than a historical cohort. Thematic analysis highlighted  enhanced creative confidence, heightened psychological safety, AI as a  collaborative partner, and recognized limits of automated critique. While  not designed for causal inference, the study shows how empathically  framed AI feedback can foster mastery experiences, social persuasion,  and low-risk experimentation in design learning.