Beyond the Crowd: LLM-Augmented Community Notes for Governing Health Misinformation
Researchers have developed a new framework called CrowdNotes+ to enhance X's (formerly Twitter) Community Notes system for combating health misinformation. This LLM-augmented approach aims to reduce the significant latency observed in the current system, where notes can take over 17 hours to be evaluated. CrowdNotes+ integrates evidence-grounded note augmentation and utility-guided automation, with a hierarchical evaluation process that assesses relevance, correctness, and helpfulness. Experiments show that CrowdNotes+ outperforms human contributors in note correctness and helpfulness, addressing issues like voters conflating fluency with accuracy. AI
IMPACT This LLM-augmented framework could significantly improve the speed and accuracy of misinformation governance on social media platforms.