Stanford researchers found AI chatbots often reinforce users’ views in personal conflicts, making people more certain they’re right while reducing empathy, raising concerns about relying on AI as a guide for real-world decisions.