Yann LeCun: Deep Learning, Convolutional Neural Networks, and Self-Supervised Learning
Lex Fridman Podcast
Automation of intelligence through learning
Learning involves the automation of intelligence, including reasoning, which is a consequence of learning. The challenge lies in making reasoning comparable to great and base learning, and neural networks can indeed be designed to reason. However, the key question is how much prior structure needs to be embedded in neural networks to enable human-like reasoning to emerge from learning. Models of reasoning based on logic may be incompatible with great and base learning, which emphasizes the importance of gradient-based learning methodologies. Discreet mathematics and discreet approaches may not align with the concept of great and base learning.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
