Machine Learning Street Talk (MLST)

#54 Gary Marcus and Luis Lamb - Neurosymbolic models

5 snips
Jun 4, 2021
In this engaging discussion, Gary Marcus, a renowned scientist and AI entrepreneur, alongside Luis Lamb, Secretary of Innovation for Science and Technology in Brazil, dive into the future of artificial intelligence. They challenge the limitations of deep learning and advocate for a hybrid neurosymbolic approach to enhance AI understanding and reasoning. Topics include the importance of integrating symbolic reasoning, exploring the complexities of abstraction, and the role of intention in knowledge acquisition. Their insights illuminate the path towards more sophisticated AI systems that can genuinely understand and reason like humans.
Ask episode
AI Snips
Chapters
Books
Transcript
Episode notes
INSIGHT

ML vs. Programming

  • Machine learning, like curve fitting, relates inputs to outputs, while programmers define algorithms independently of data.
  • This difference impacts how these systems generalize and handle novel situations.
INSIGHT

Deep Learning Limitations

  • Gary Marcus's previously criticized views on deep learning's limitations are now widely accepted.
  • Even his critics like Yann LeCun now acknowledge issues with semantics and generalization.
ANECDOTE

Marcus vs. Bengio Debate

  • Gary Marcus and Yoshua Bengio debated symbolic vs. connectionist approaches.
  • They disagreed on the core representations and how to integrate different AI approaches.
Get the Snipd Podcast app to discover more snips from this episode
Get the app