AI chats reduce conspiracy belief

A large experiment (n=2,190) found that three‑round GPT‑4 conversations tailored to a person’s own conspiracy reduced their belief by about 20%, with effects persisting at least two months. A professional fact‑checker rated 99.2% of the AI’s sampled claims true and none false, and reductions spilled over to unrelated conspiracies. — This suggests AI could be deployed as a scalable debunking tool, reframing policy from AI as a disinfo threat to AI as a potential public‑interest 'engine of truth.'