AI Research
AI's Step-by-Step Lies: Chain-of-Thought's Dirty Secret
Chain-of-Thought reasoning was supposed to make AI transparent. Turns out, it's often just post-hoc BS from models that already know their answer.