The problem with AI dialogue at scale | Science

In their Research Article “Durably reducing conspiracy beliefs through dialogues with AI” (13 September 2024, p. 1183), T. H. Costello et al. provide evidence that dialogue with generative artificial intelligence (AI) can reduce conspiracy beliefs. The paper’s conclusions frame AI as a potentially scalable intervention that can save societies from misinformation and conspiracy theories (1, 2). However, the applicability of this strategy to real-world problems remains unclear.