How Brains Can Still Outperform our Best Computing Machines

Our meetings are hybrid; you may attend in-person or on Zoom. Register here to get information about attending.

Conventional computers have grown exponentially more powerful in recent decades, but human-made programs still struggle with many hard problems. The growth of neuro-inspired AI has unlocked new capabilities for some hard problems, such as image identification and natural language processing. Fundamental work in this area won a Nobel Prize just recently.

Still, there are several domains in which brains still outperform our best computing machines. One example is visual navigation, as self-driving cars–even with a broad array of cameras, sensors, and computation–still struggle to reliably navigate. Our speaker will discuss these hard problems and some ideas about why brains are so effective at some hard tasks, including ideas for improved computation for hard problems.

Dr. Spencer LaVere Smith, Associate Professor, UCSB Electrical & Computer Engineering
Dr. Smith investigates computation in neurobiology and optics. His lab (slslab.org) builds multiphoton imaging technology for measuring and manipulating neural activity and performs experiments to gain mechanistic insights into neural circuitry. The lab also investigates machine learning and optical computing. He is also the CEO of Pacific Optica, a startup focused on optics that unlock new possibilities.
Dr. Smith earned his undergraduate degree in physics and mathematics at the University of Iowa and his PhD in neuroscience and neuroengineering at UCLA. After postdoctoral work at University College London, he moved to UCSB in 2018. His awards include a PECASE (2019) and a McKnight Technological Innovation Award (2015).
Location: Our speaker will present in person at Community West Bank conference room at 445 Pine Ave. off Hollister Ave. in Goleta, and on Zoom. When you register, we send a confirmation email containing Zoom data, directions to the in-person location and meal details.