Automated Alignment Researchers: Using large language models to scale scalable oversight - Anthropic

Automated Alignment Researchers: Using large language models to scale scalable oversight Anthropic