
On January 20, 2026, the newly created Center of Excellence for AI and Robotics held the first Hallam Winter Seminar on Intelligence, bringing together researchers and students from universities and companies across Yorkshire to discuss current topics in natural and artificial intelligence.
In this inaugural edition, the central theme was “The Active Inference Pathway to Alignment,” featuring talks by researchers from Canada, Oxford, and London. Active inference is a neuroscience-inspired framework for embodied intelligence that integrates perception and action, seeking to minimize an agent’s surprise about its environment. The framework is increasingly being used to redefine the issue of alignment as a problem of mutual inference, with failures understood as breakdowns in its various components—particularly when information is incomplete or the generative model is misleading.
The event concluded with a panel discussion on pressing issues related to AI alignment in the modern world. Events of this kind are key to incentivizing and stimulating collaborative thinking, bringing together academics with diverse expertise.