Programme

The detailed programme, with times and talks, is coming soon.

Session 1 · Energy
Brains achieve remarkable computational performance under severe energetic constraints: neuronal signalling, synaptic transmission and plasticity are all metabolically costly, yet biological systems remain vastly more energy-efficient than today’s large artificial neural networks. Understanding how nervous systems trade off accuracy, speed, robustness and energy consumption can therefore inform more sustainable AI. This session will bring together experimental neuroscience perspectives on the energy cost of neural computation with AI and industry approaches to measuring, reducing and governing the environmental footprint of machine learning.
Session 2 · Learning
Biological neural circuits learn continuously, from sparse data, using local and energy-cheap plasticity rules—a stark contrast to the massive labelled datasets, centralised gradients, and repeated retraining that conventional deep learning relies on. This session brings together neuroscientists developing biologically plausible learning algorithms with computer scientists designing new ways to make training and inference more efficient, from distillation and pruning to quantisation. Together, they will ask what principles—biological or algorithmic—can make learning itself cheaper, faster, and more robust.
Session 3 · Representation
The brain encodes the world in compact, structured, and reusable representations—sparse codes, low-dimensional manifolds, composable features—that generalise far beyond the situations in which they were learned. AI systems, by contrast, tend to build high-dimensional, redundant, and brittle representations that transfer poorly and require enormous parameter counts to store what the brain holds in a fraction of the space. The session asks what principles govern the geometry and structure of biological representations, why they are so much more efficient than artificial ones, and whether these principles can be reverse-engineered into AI systems.
Session 4 · Roundtable
The final session steps back from the technical questions for a more philosophical, debate-driven discussion of what efficiency means for brains, machines and the science of intelligence: a roundtable debate with the day’s speakers.