Matthew Kenney – Language Model Capabilities Benchmarking

[Navigating Transformative AI] Open Philanthropy recommended a grant of $397,350 to Matthew Kenney to support a project developing evaluations that will benchmark the ability of language model agents to accelerate AI R&D. This grant was funded via a request for proposals for projects benchmarking LLM agents on consequential real-world tasks. This falls within our focus area of potential risks from advanced artificial intelligence.