Meeting Reports: FENS Forum 2022

Summary

In April 2022, DALL-E 2 was launched to immense popularity: it is an artificial intelligence (AI) model that creates an image corresponding to a phrase that the user provides. Prompted with phrases like “impressionist painting of a robot neuroscientist” or “a dragon perched on a snowy mountain,” DALL-E 2 creates realistic images with interesting compositionality. Notably, these features were not explicitly designed into the generative machine learning model, and these emergent properties have been hitherto uniquely attributed to human creativity. This raises the question of what differentiates human, and more broadly, biological intelligence, from AI technologies like DALL-E 2, AlphaGo, and others? What do these similarities or differences tell us about the nature of intelligence, and how can the two communities of AI and neuroscience (which studies biological intelligence) benefit from understanding how artificial and biological intelligence intersect with each other?

These questions formed the basis of the Building and understanding brains: How can AI research inform neuroscience? debate held on July 9, 2022, as part of the Federation of European Neuroscience Societies (FENS) conference in Paris. The debate was moderated by Christopher Summerfield, Professor at the University of Oxford, and featured prominent panellists, in order of their presentations: Jane Wang, Staff Research Scientist at DeepMind; Kanaka Rajan, Assistant Professor at Mount Sinai; Claudia Clopath, Professor at Imperial College London; Andrew Saxe, Sir Henry Dale Fellow and Associate Professor at University College London; and Stanislas Dehaene, Director of NeuroSpin and professor at College de France.

Read more in PDF