Introducing LifeSciBench

OpenAI has released LifeSciBench, a collection of expert-evaluated tasks designed to measure the proficiency of AI models in life science research and decision-making. By testing systems against real-world scientific challenges rather than general knowledge, the benchmark aims to provide a more accurate assessment of how AI can assist in laboratory work and professional biological research.