Paper Explainer

Can an LLM Learn to Be Honest About What It Doesn't Know?

Reinforcement Learning with Metacognitive Feedback trains a model not just to answer well, but to judge itself — and it turns out that models rewarded for calibrated self-assessment get better at tasks and express faithful uncertainty. Based on the paper shared by pybeebee.

The Training Pipeline

The blueprint from the paper's diagram: a query flows through the policy model, its answer is scored on two axes — task success and metacognitive accuracy — and both rewards drive the RL update. Tap each stage to expand it, or press Play.

Reward Simulator

You are the reward function. Pick whether the model's sampled answer is actually correct, then drag its stated confidence. Watch the task reward and the metacognitive (calibration) reward — and see which behavior the RL update reinforces.

Why Calibration Emerges

The metacognitive reward peaks when stated confidence equals actual accuracy. Overclaiming on wrong answers is punished hardest — exactly the hallucination-with-bravado failure mode. The chart shows reward as a function of confidence for correct vs. wrong answers.

The dot tracks your simulator settings above. The honest policy is a truthful map from internal knowledge to expressed confidence — that is faithful uncertainty expression.

Key Takeaways

Reinforcement learningMetacognitionCalibrationHonesty in LLMsUncertainty quantification
Enjoy this tool? Build your own with Super