Model2h ago
Model
Chatbots Lean Left On Political Queries
A Washington Post analysis found that several popular AI chatbots produce more left-leaning responses than right-leaning ones when answering political questions. The study highlights persistent alignment biases in large language models. This suggests that RLHF processes often bake in specific ideological leanings, complicating the pursuit of truly neutral AI assistants.