When you don’t know an answer, asking an LLM for an explanation feels productive. Cognitive science says the opposite of what feels easy: trying to recall first, then getting feedback, is far more effective for retention.
Evidence
The testing effect (retrieval practice) is one of the most robust findings in learning science. Across hundreds of studies, testing improves long-term retention more than restudying the same material. Recent work comparing AI explanations to retrieval practice finds that students who attempt recall before reading explanations retain more after a delay.
| Strategy | Immediate performance | Delayed retention (7–14 days) | Notes |
|---|---|---|---|
| Restudy (reread notes/explanation) | High | Low | Illusion of mastery |
| Retrieval practice (test yourself) | Moderate | Highest | Desirable difficulty |
| AI explanation only | High (feels fluent) | Moderate–low | Passive consumption |
Crucially, the benefit comes from the retrieval attempt, even if you fail. The failure triggers better encoding of the correct answer when you see it.
Practical application with AI
Instead of “explain this”, use a test-first workflow:
- Attempt to answer from memory without prompts.
- Only then ask the AI for a targeted explanation of your mistake.
- Close the chat and retrieve again later (spaced retrieval).
- Use the AI to generate practice questions, not just answers.
This preserves the retrieval effort while still getting precise feedback.
VERDICT
Use AI to create practice tests and to explain specific errors after retrieval. Don't use it to pre-chew answers before you've tried to recall them.
Sources
- Roediger and Karpicke, “Test-enhanced learning: Taking memory tests improves long-term retention”, Psychological Science 17, 2006
- Karpicke and Roediger, “The critical importance of retrieval for learning”, Science 319, 2008
- Adesope et al., “Rethinking the use of tests: A meta-analysis of practice testing”, Review of Educational Research 87, 2017