Yoshua Bengio: From System 1 Deep Learning to System 2 Deep Learning | NeurIPS 2019


Runs 55:02

A talk by Turing Award recipient Yoshua Bengio to an international assembly of AI researchers. Bengio is one of three Turing winners for 2019, named with Geoffrey Hinton and Yann LeCun for their pioneering work in deep learning. He outlines a research path towards human-level AI.

Bengio connects present and future states of deep learning to concepts of cognition described in ‘Thinking fast and Slow,’ a book by Daniel Kahneman. He says ‘System 1’ thinking is like current deep learning:

  • Intuitive
  • Fast
  • Unconscious
  • Non-linguistic
  • Habitual

Future deep learning would be like ‘System 2’:

  • Slow
  • Logical
  • Sequential
  • Conscious
  • Linguistic
  • Algorithmic
  • Planning
  • Reasoning

Bengio goes on to frame the missing pieces and how a research agenda can can link one to the next.

His talk was given at NeurIPS 2019, the world’s largest conference for AI specialists. It is intended for a computer science audience. This version combines his slides and address in the same frame.

VANCOUVER | December, 2019

Go to journalismAI.com homepage

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.