AI RESEARCH
PaperBench: Evaluating AI’s Ability to Replicate AI Research
OpenAI Blog
•
We introduce PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research.