
AI Frontiers: A deep dive into deep learning with Ashley Llorens and Chris Bishop
Microsoft Research Podcast
The Future Integration of Language Models and Deep Learning in Scientific Discovery
Scientists envision a future where language models with an understanding of molecular structures and scientific literature will aid in scientific discovery. However, it is uncertain whether all tools will be integrated into one overarching tool. Deep learning is currently revolutionizing scientific discovery, particularly in simulating fundamental equations of nature. This transformation leverages architectures like diffusion models, large transformers, and large GPU clusters, but the techniques used differ from those in processing natural language. In the future, scientists may have a super co-pilot that can be interacted with in natural language and draw upon various tools for assistance.