Cambridge researchers are advancing artificial intelligence applications in various scientific fields after being awarded new Encode: AI for Science Fellowships. These fellowships place leading AI specialists in prominent UK laboratories to address complex scientific problems and accelerate the development of practical solutions. The initial group includes three fellows based at Cambridge.
Jonathan Carter, one of the Encode Fellows, is adapting technology from astrophysics to explore how humans intuitively understand physics. Working with Hiranya Peiris, Professor of Astrophysics at Cambridge, Carter uses interpretable variational encoders—a type of neural network that identifies meaningful patterns in complex data. Their research aims to improve understanding of human intelligence and enhance AI systems' ability to learn and generalize similarly to humans.
Shruti Mishra, another fellow, is developing an AI system designed to generate understandable equations describing turbulent flows across different scales. This issue has implications for weather forecasting and aerospace engineering. Under the guidance of Miles Cranmer, Assistant Professor of Data Intensive Science at Cambridge, Mishra combines machine learning with symbolic mathematics so scientists can interpret the resulting equations rather than rely on opaque predictions.
Martyna Stachaczyk is collaborating with Rika Antonova, Associate Professor at Cambridge, on a biologically inspired control architecture for real-time local intelligence in devices. Their work could enable intelligent systems—such as prosthetics or environmental sensors—to operate independently from cloud computing infrastructure, which may be unavailable or insecure in some settings.
The Encode AI for Science Fellowship program is managed by Pillar VC and receives funding from the Advanced Research + Invention Agency (ARIA) as well as the UK Government’s Sovereign AI Unit.