Biologically-inspired Artificial Intelligence:
Challenges and Opportunities

Artificial intelligence has become a transformative force in scientific research and society at large, yet current models face significant limitations. Transformer-based large language models, despite their impressive capabilities, suffer from excessive energy consumption, data requirements, and concerning issues like hallucinations. As the performance gains of these architectures begin to plateau, there is growing recognition that the path toward more capable and efficient AI may require fundamentally new approaches. One of the most promising directions is to take inspiration from biological brains, intelligent systems that have been refined through millions of years of evolution.

This workshop at Cosyne explores how brain-inspired principles can address current challenges in artificial intelligence. From neuromorphic computing to biologically-plausible learning rules, and from sleep mechanisms to neuromodulation, recent advances demonstrate the breadth of insights that neuroscience can offer to AI development. Our speakers will present work highlighting both theoretical foundations and practical applications, examining how biological principles can inform the development of more efficient and capable AI systems and identifying promising research directions at this interdisciplinary boundary.

Alongside an exciting list of talks we will organise a specific NeuroAI posters session during the lunch break (see the form below) and are also working on organising extra sessions for Early Career Researchers to learn more about industry careers in NeuroAI. We are excited to welcome the following speakers to our workshop:

• Jonathan Cornford (University of Leeds) - What can we learn about optimisation from biological circuits? 

• Ida Momennejad (Microsoft Research) - Brain-inspired memory and reasoning

• Yulia Sandamirskaya (Zurich University of Applied Sciences) - Brain circuits for autonomy: how will next-get AI look like

• Wolfgang Maass (TU Graz) - Episodic memories make goal-directed action selection context-aware and explainable

• Shahab Bakhtiari (University of Montreal) - Sequential Predictive Processing and Efference Copy: Cortical Mechanisms for World Modeling

• Filippo Moro (Institute of Neuroinformatics, University of Zurich and ETH Zurich) - Three Levels of Biological Inspiration for Neuromorphic Computing and NeuroAI

• Charlotte Frenkel (TU Delft) - A NeuroAI Framework for Intelligent Systems

• Rui Ponte Costa (University of Oxford) - NeuroAI models of learning: subcortex, cortex and neuromodulation

The workshop is organized by Nicolas Skatchkovsky (nicolas.skatchkovsky@crick.ac.uk) and Jascha Achterberg (jascha.achterberg@dpag.ox.ac.uk). Please reach out to us with any questions and suggestions!