To many AI researchers it came as a surprise that a large neural network trained only to predict the next word can show signs of general intelligence. Neuroscientists who hold the “predictive brain” hypothesis — that brains evolved above all to anticipate the future — were less surprised.  In this talk, based on his book of the same title, Blaise Agüera y Arcas takes this idea further and argues that prediction is fundamental not just to brains but to life itself.  He offers a broad picture of intelligence that runs from molecules to organisms, societies, and AI, drawing on computer science, biology, physics, and neuroscience.

Reception at 5:30 PM, talk at 6 PM.
Please register in advance if you plan to attend.

__

To request the Zoom link send an email to jteeters@berkeley.edu. Also indicate if you would like to be added to the Redwood Seminar mailing list.

—

To request the Zoom link send an email to jteeters@berkeley.edu.  Also indicate if you would like to be added to the Redwood Seminar mailing list.

Today’s AI systems are powerful, but they work on principles very different from those of the human brain. They typically learn from large, static datasets, operate as black boxes, and need retraining for each new task. The brain, by contrast, learns continuously through active interaction with the world, builds structured models of what it senses, and generalizes across many kinds of tasks with remarkable efficiency.  In this talk, Jeff Hawkins explains how the brain’s sensorimotor principles allow it to understand the world, and why reverse-engineering the neocortex may provide a blueprint for a new kind of artificial intelligence.  The Thousand Brains Project is an open-source effort to turn these principles into a practical framework for researchers and developers.

Reception at 5:30pm, talk at 6pm.
Please register in advance if you plan to attend.

__

To request the Zoom link send an email to jteeters@berkeley.edu. Also indicate if you would like to be added to the Redwood Seminar mailing list.