Rethinking AI-assisted writing instruction: feedback literacy scripts, calibration training, and student writing...

As generative AI becomes increasingly integrated into writing instruction, the central educational challenge is not only to provide feedback but also to help students interpret, evaluate, and use such feedback critically. This study examined whether two metacognitive interventions-a Feedback Literacy Script (FRAC) and an Assessment-Performance Calibration Activity (APCA)-could improve students' writing quality and self-assessment accuracy in AI-assisted writing.

A 2 × 2 mixed factorial design was employed with 120 undergraduate English majors assigned to four conditions: regular AI use, FRAC only, APCA only, and FRAC+APCA. Across four writing task points, the study collected data on writing quality gain, self-assessment accuracy, overconfidence, effective feedback uptake, and revision depth.

The findings showed differentiated effects of the two interventions. FRAC had a stronger direct effect on writing quality, effective feedback uptake, and deep revision, suggesting that feedback literacy training primarily improved how students processed and enacted AI feedback. APCA showed the clearest effect on self-assessment accuracy and overconfidence reduction, indicating that calibration training more directly strengthened students' internal judgment. The combined intervention produced the highest gains in overall writing quality and the strongest retention after AI support was removed, but it did not outperform APCA alone in self-assessment accuracy.

These findings indicate that better writing and more accurate self-evaluation are related but distinct outcomes in AI-assisted writing. The study suggests that the educational value of AI-assisted writing depends less on feedback abundance itself than on whether learners can evaluate feedback, calibrate self-judgment, and transform external support into independent revision ability.