Tag
This paper investigates how prompt robustness varies between objective and subjective questions in LLM evaluations, finding that sensitivity to prompt changes depends on question type, prompt change, and model.
OpenAI presents a comprehensive framework for defining and evaluating political bias in LLMs, introducing a 500-prompt evaluation spanning 100 topics across five bias axes. Results show GPT-5 models achieve 30% bias reduction compared to prior versions, with less than 0.01% of production ChatGPT responses exhibiting political bias.
OpenAI publishes a blog post outlining its commitment to intellectual freedom in ChatGPT design, emphasizing objectivity by default, user controls, and transparent principles through its Model Spec framework. The company highlights new personalization settings and ongoing efforts to evaluate and reduce political bias through stakeholder feedback.