2024-11-28 08:00 – Ištrauka iš laiško „Aida Daily: NASA's Cold War Discovery / China's Defense Shake-up / Bluesky's User Surge"
AI Tops Experts in Neuroscience Predictions
UCL researchers have unveiled BrainBench, a new benchmark showing large language models (LLMs) outperform human experts in neuroscience predictions with 81% accuracy versus 63%. Even better, their specialized BrainGPT hits 86%. By analyzing complex abstracts, these AI models promise to turbocharge scientific discovery. As LLMs integrate seamlessly into research workflows, could your next breakthrough be co-authored by an AI? Stay tuned as AI takes a deeper dive into the brain!
