Research at the frontier of understanding how large language models process and respond to political content — treating AI systems as subjects of inquiry rather than just tools.
Political Sycophancy
In Prep · LLM evaluation, mechanistic interpretability
Do language models adjust their political responses based on perceived user identity?