When adult learners complete AI‑supported multiple‑choice quizzes that report separate independent and hint‑assisted scores, does adding a brief, structured post‑session reflection (where learners predict their no‑hint performance next week) further reduce illusions of learning and overconfidence compared with score separation alone, especially for low‑prior‑knowledge learners?

Answer

Adding a brief, structured post‑session reflection that asks learners to (a) predict their no‑hint performance on a similar quiz next week and (b) justify that prediction is likely to further reduce illusions of learning and overconfidence beyond separate independent vs. hint‑assisted scores alone, with the strongest incremental benefit for low‑prior‑knowledge learners—provided the reflection is specific, required, and periodically checked against actual delayed, no‑hint performance. The effect will be smaller or unreliable if predictions are never benchmarked against delayed tests, if the prompts are vague, or if learners can skip the reflection with little friction.