Long Phan, Devin Kim, Alexander Pan +3
Large language models (LLMs) exhibit systematic political bias across a variety of sensitive contexts. We find that LLMs handle counterpart topics from opposing political sides asy…