OpenAI announces “HealthBench” to evaluate medical AI models. LLM > specialist doctors, but there is almost no difference in scores between “LLM alone” and “doctor + LLM”...

This week in medical news, OpenAI announced HealthBench, a new way to evaluate medical AI models that suggests large language models (LLMs) may soon surpass specialists, though LLM-supported doctors currently perform similarly to standalone LLMs. In another development, AI Scientist, a multi-agent system, discovered a promising drug candidate for a major cause of blindness, demonstrating a closed-loop AI approach to scientific discovery. Simultaneously, a "Don't Die" movement is gaining traction in Silicon Valley focused on radical life extension and utilizing services like genetic testing kits, while traditional weight loss programs are facing challenges with the bankruptcy of WW International amidst the rise of GLP-1 medications.