Scientific foundation models for genetic disease research
Scientists are creating scientific foundation models, a type of large language model trained on natural data, such as proteins, DNA, and chemicals. These models are able to learn the language of life and can be used to create novel molecular compounds of enormous medical potential. And they can run on a single GPU rather than a large-scale institutional computing cluster, which greatly expands the universe of researchers who can access them. AI scientist Stanley Bishop will discuss the impact scientific foundation models could have on our medical system, with focus on rare genetic disease and oncology research.