THE IMPACT OF AI-BASED FEEDBACK ON METACOGNITIVE SKILLS OF UNIVERSITY STUDENTS: A CROSS-SECTIONAL ANALYSIS

Artificial Intelligence (AI) has begun to reshape feedback in higher education, but its impact on metacognitive skills is not well understood. This mixed-method approach, using a quasi-experimental design, assessed the effects of introducing an adapted pedagogy with an AI-based feedback system (ChatGPT-4) on 120 first-year university students. Two studies were conducted: a tenth-grade EG (n=60) was assigned personalized AI feedback for eight weeks, and a tenth-grade CG (n=60) was assigned traditional teacher feedback for eight weeks. The Metacognitive Awareness Inventory (MAI, α=0.91) was used as a pre- and post-test instrument and combined with focus groups and learning diaries. The quantitative results indicated statistically significant differences (with the number of respondents in favor of the EG in the respective domain) between the EG and CT in planning (p<0.001, d=1.12) and monitoring (p<0.001, d=0.96), but not in evaluation (p=0.245). A mixed ANOVA revealed a significant time×group interaction (F=42.31, p<0.001, partial η²=0.26). Three themes emerged from the thematic analysis: (1) the need for prompt feedback from AI raises awareness of the need for online monitoring, (2) metacognitive questioning supports strategic planning but potentially heightens reliance on technology, and (3) perceived usefulness is dependent on the type of task. Conclusions: The feedback provided by AI in the form of metacognitive skills has a significant impact on planning and monitoring skills but does not positively influence metacognitive evaluation. The mixed-methods approach found that immediacy activates self-regulatory processes in students, yet it can also create dependency that can be compensated by teacher-guided reflections. HLEIs are encouraged to start using AI systems with explicit metacognitive scaffolding and AI activities in HE.