PROJECT LIGHTHOUSE
Ethan Mollick: The next generation of AI models (post-2025) will outperform human experts on average on the METR long-task agentic benchmark. — Project Lighthouse