Sutton’s “dead end” argument for LLMs
01 October 2025

Sutton’s “dead end” argument for LLMs

AI IN 5,4,3,2,1... (english)

About
In this episode of "AI in 5,4,3,2,1," Dominic, founder of 'Leaders of AI,' shines a light on an intriguing controversy in the world of Artificial Intelligence: Are large language models the right path or merely imitators of human knowledge? This discussion leads us into new horizons for learning in AI.

- Richard Sutton, a pioneer of reinforcement learning, questions the scaling of large language models.
- Comparison with a chef: Theory vs. practical experience in reality.
- Introduction of a new approach called "Oak," an autonomous, self-learning agent.
- Paradigm shift: From rigid models to dynamic, continuously learning systems.

More info can be found at: https://www.youtube.com/live/XqYTQfQeMrE?t=22620s