MIT researchers across nearly every department are leveraging AI — including machine learning, large language models, and neural networks — to accelerate discovery in drug development, materials science, energy, robotics, neuroscience, and ecology. Highlights include the Boltz protein-structure prediction models (culminating in BoltzGen for custom protein design), autonomous AI-driven robotic labs testing hundreds of polymer blends per day, AI-optimized jet turbine components, and power grid scheduling tools. Researchers also candidly discuss limitations: AI hasn't yet cracked neurodegenerative disease mechanisms, physics simulators fail to capture real-world manufacturing complexity, and hallucinations remain a concern. The overarching view is that AI has already consumed the low-hanging fruit and the harder, higher-impact challenges lie ahead.
Sort: