Short Dialogues with AI Reduce Belief in Antisemitic Conspiracy Theories

"- Antisemitic conspiracy beliefs are common and closely linked to negative attitudes toward Jews. Participants who endorsed more antisemitic conspiracy theories consistently expressed less favorable views of Jewish people. - A brief dialogue with an AI significantly reduced belief in these conspiracies. Participants who engaged in a targeted, fact-based conversation with Claude 3.5 showed notably larger drops in belief than those who received only a label warning or who discussed an unrelated topic. - The effect was broad and reliable. Reductions in belief appeared across all antisemitic conspiracy theories tested and held up under multiple robustness checks. - Among participants who initially held unfavorable views of Jews, attitudes improved. The AI dialogue increased favorability toward Jews within this subgroup, indicating that correcting specific misconceptions can meaningfully influence broader attitudes. - The changes showed partial durability. When participants were surveyed again about a month later, roughly half of the initial reductions in belief—and about half of the improvement in attitudes among initially unfavorable participants—remained. - Bottom line: Brief, fact-focused conversations with an AI can meaningfully reduce belief in antisemitic conspiracies and, for those starting out most negative, can also soften antisemitic attitudes."