PoliticsBench: Benchmarking Political Values in Large Language Models with Multi-Turn Roleplay

PoliticsBench: Benchmarking Political Values in Large Language Models with Multi-Turn Roleplay explores A new benchmark and framework to evaluate and quantify political bias in large language models through multi-turn roleplay.. Commercial viability score: 4/10 in LLM Bias Evaluation.