The Model Cannot Mark Its Own Work
中文摘要
AI模型无法客观地批改自己的答案，即便有自我修正的痕迹，依然是自评行为。
English Summary
AI models cannot objectively grade their own work; sequences showing doubt and self-correction are still just self-marking.
原文节选
A trace that shows sequence, doubt and self-correction can still be the model marking its own exam. · Architecting the AI Coworker · 7/22 Continue reading on Medium »