Gruene

LLMs show centrist, status-quo bias on concrete policy votes

A new study published on arXiv challenges the notion that large language models (LLMs) exhibit a consistent left-leaning political bias. Researchers found that while LLMs align with established findings when answering a…