I-CALM: Incentivizing Confidence-Aware Abstention for LLM Hallucination Mitigation

Researchers developed I-CALM, a prompt-based framework that reduces AI hallucinations by encouraging language models to abstain from answering when uncertain, rather than providing confident but incorrect responses. The method uses verbal confidence assessment and reward schemes to improve reliability without model retraining.

• →I-CALM framework reduces false answer rates by encouraging AI models to abstain when uncertain rather than hallucinate responses.
• →The method works through prompt-only interventions without requiring model retraining or modification.
• →Results show a clear trade-off between coverage and reliability - fewer answers but higher accuracy on answered questions.
• →Self-reported verbal confidence serves as a stable and well-calibrated uncertainty signal for language models.
• →The framework incorporates normative principles emphasizing truthfulness, humility, and responsibility in AI responses.