x
The Evolution of Agentic AI Evaluation — LessWrong