AI
Anthropic's Approach to Metacognition in Large Language Models
Anthropic researchers are applying metacognition concepts to improve LLM reliability by teaching models to identify their own knowledge gaps and blind spots before executing tasks.
September 23, 2026, 08:10 AM·Хабр
