Richard Sutton on Dwarkesh Patel: Why LLMs Lack a Goal

Name: Richard Sutton on Dwarkesh Patel: Why LLMs Lack a Goal
Uploaded: 2025-09-26T00:00:00Z
Duration: 1 h 7 min 8 s
Description: Richard Sutton argues that large language models (LLMs) are fundamentally limited because they imitate human text without grounding in experience, goals, or consequences, whereas reinforcement learning (RL) is built around agents acting in the world, getting reward, and learning from outcomes over time.

How temporal difference learning gives AI a ground truth that LLMs lack: Sutton argues without reward signals, there is no right or wrong action.

Richard SuttonguestDwarkesh Patelhost

Sep 26, 20251h 7mWatch on YouTube ↗

CHAPTERS

0:00 – 13:51
Are LLMs a dead end?
13:51 – 23:57
Do humans do imitation learning?
23:57 – 34:25
The Era of Experience
34:25 – 42:17
Current architectures generalize poorly out of distribution
42:17 – 47:28
Surprises in the AI field
47:28 – 54:35
Will The Bitter Lesson still apply after AGI?
54:35 – 1:07:08
Succession to AI

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome

Are LLMs a dead end?

Do humans do imitation learning?

The Era of Experience

Current architectures generalize poorly out of distribution

Surprises in the AI field

Will The Bitter Lesson still apply after AGI?

Succession to AI

Get more out of YouTube videos.