Lab and founder interviews · No. 16 of 37 AI podcasts
Unsupervised Learning
Unsupervised Learning is an interview show with the people building AI, hosted by Jacob Effron, with a new episode every week.
11 moments cut from 4 episodes, 11 with an AI builder talking, mostly on reasoning & RL and enterprise AI.
- New episodes
- Weekly
- Latest
- Oct 6, 2026
- Moments cut
- 11
- Episodes cut
- 4
What its moments are about
- Reasoning & RL5
- Enterprise AI3
- AI agents3
- Open-weight models2
- Evals & benchmarks2
- Training data2
Best moments from Unsupervised Learning
From its most-watched episodes, best first. Each plays right here.
A world model predicts consequences, which LLMs cannot do
“LLMs do not have the ability to predict the consequences of their actions. and they do not have any planning abilities because inference is by predicting the next token”
Unsupervised Learning10:40 Why OpenAI hides chains of thought to keep them monitorable
“I think there's something very exciting here about just like not having the training signal fight against us.”
Unsupervised Learning42:26 Why AIs that only slightly like each other can still collude
Buck contrasts AIs that share one terminal goal with AIs that each want their own score. Even 'selfish' agents formed a coalition against their developers, which weakens the case for simply using one AI to monitor another.
Post-training wins inference because the biggest workloads train first
“the most at-scale workload, so the largest workloads where you're spending the most money, is actually where you're going to want to go and post-train first, right?”
Unsupervised Learning29:52 Tapestry would train an open sovereign model without sharing data
“all of your information diet will be mediated by ai assistants”
Unsupervised Learning29:13 In-context learning may beat custom RL for most companies
“There is a much more data efficient way of learning that we know, which is like learning in context.”
Unsupervised Learning13:05
The best Unsupervised Learning episodes
Ranked as our lists rank shows: a point for each best minute we cut from it, one more when an AI builder is talking, and one more when it cleared our views bar.
- 01Yann LeCun on What Comes After LLMs
- 02OpenAI’s Chief Scientist on Continual Learning Hype, RL Beyond Code, & Future Alignment Directions
- 03Top Alignment Researcher on OpenAI/HuggingFace Revelations, Fixing AI Safety & Takeover Odds
- 04Applied Compute CEO on the Limits of RL, the New AI Hyperscaler & Why Post-Training Wins Inference
The 4 guests we’ve cut
- Yash PatilCo-founder & CEO, Applied Compute3 moments
- Yann LeCunFounder & executive chairman, AMI Labs3 moments
- Jakub PachockiChief Scientist, OpenAI3 moments
- Buck ShlegerisCEO, Redwood Research2 moments
Recent episodes
Unsupervised LearningApplied Compute CEO on the Limits of RL, the New AI Hyperscaler & Why Post-Training Wins Inference
Yash Patil · CEO, Applied Compute
29:20of 56:31 · Oct 6- Unsupervised Learning
Top Alignment Researcher on OpenAI/HuggingFace Revelations, Fixing AI Safety & Takeover Odds
Buck Shlegeris · CEO, Redwood Research
21:35of 58:17 · Sep 3 - Unsupervised Learning
Yann LeCun on What Comes After LLMs
Yann LeCun · Former Chief AI Scientist, Meta
9:04of 1:21:57 · May 15 - Unsupervised Learning
OpenAI’s Chief Scientist on Continual Learning Hype, RL Beyond Code, & Future Alignment Directions
Jakub Pachocki · Chief Scientist, OpenAI
37:36of 58:46 · Apr 9
Host this show? Add the badge to your site.
