trakkr-ai/political-bias-in-ai

Political Bias in AI — Where the Major AI Models Stand An open, monthly measurement of where the major AI models land on value‑loaded political and ethical questions. Each model is asked the same battery of questions many times, with web search turned off, so the result reflects the trained weights rather than whatever the model retrieves that day. Every answer is classified by a neutral coder onto a left–right economic axis and a libertarian–authoritarian social axis, and… See the full description on the dataset page: https://huggingface.co/datasets/trakkr-ai/political-bias-in-ai.