@zihaogavinyang
AI
The paper’s author notes that paraphrasing often changes an LLM’s answer too, raising questions about the baseline for bias tests.
AI
