The new Fellowships aim to place top AI talent in the UK’s leading labs, addressing major scientific challenges and accelerating real-world applications. Three Fellows from the first cohort are currently based at Cambridge.

Jonathan Carter, an Encode Fellow, is adapting technology first developed for astrophysics to explore how humans intuitively understand physics — such as predicting the path of a thrown ball. Working with Hiranya Peiris, Cambridge Professor of Astrophysics (1909), Carter employs interpretable variational encoders, a type of neural network that uncovers compact, meaningful representations within complex data. This cross-disciplinary research could deepen our understanding of human intelligence while advancing AI systems that learn and generalize more like people.

Shruti Mishra, another Encode Fellow, is developing AI methods to uncover clear, interpretable equations that explain turbulent flows across scales — a long-standing challenge in physics with wide-ranging implications, from weather forecasting to aerospace engineering. Under the guidance of Miles Cranmer, Assistant Professor of Data Intensive Science at Cambridge, Mishra combines machine learning with symbolic mathematics to generate equations that scientists can interpret and trust, avoiding the limitations of ‘black-box’ predictions. This work could significantly improve climate models and industrial design.

Martyna Stachaczyk, also an Encode Fellow, is collaborating with Rika Antonova, Associate Professor at Cambridge, to create biologically inspired, on-device control systems for real-time, local intelligence. Their goal is to reduce dependence on cloud computing — often insecure or unavailable in low-connectivity environments — and instead enable robust, adaptive autonomy in prosthetics, robotics, and environmental platforms, even under resource constraints.

Source: https://www.cam.ac.uk/news/new-encode-fellowships-boost-ai-research-at-cambridge