Home Speakers John Schulman Quote Basically anything that can be learned through RL can be distilled very easily, because it’s a small number of bits.
Words found on the source page on Nov 6, 2026.
More from John Schulman Podcast Dwarkesh Podcast: AI researchers debate how close we are to recursive self-improvement Sep 11, 2026
In fact, I remember in the early OpenAI days having the intuition that just minimizing log loss wasn’t going to get you to intelligence. Podcast Dwarkesh Podcast: AI researchers debate how close we are to recursive self-improvement Sep 11, 2026
Right now, you don’t get explosive growth in capabilities because you still get bottlenecked enough when you’re trying to do research and engineering. Podcast Dwarkesh Podcast: AI researchers debate how close we are to recursive self-improvement Sep 11, 2026
So I would expect that we’re nowhere near the ceiling of how well you can do research. Podcast Dwarkesh Podcast: Reasoning, RLHF, and Plan for 2027 AGI May 15, 2024
Even in one or two years, we'll find that the models can do a lot more involved tasks than they can do now. Podcast Dwarkesh Podcast: Reasoning, RLHF, and Plan for 2027 AGI May 15, 2024
Given the relatively small number of entities who could train the largest models, it does seem possible to coordinate. Podcast Dwarkesh Podcast: Reasoning, RLHF, and Plan for 2027 AGI May 15, 2024
If AGI came way sooner than expected we would definitely want to be careful about it.
On the same topics Blog The Specter Of Neuralese Sep 24, 2026
An AI with more real layers can actually be “smarter” in the conventional sense of the term. Blog The Specter Of Neuralese Sep 24, 2026
For 100 steps in a row, the AI can shoot delicate subtle ideas from layer to layer at the speed of light. Then there’s one step where it has to encode them into twenty-six glyphs invented by Phoenician turquoise miners in 1800 BC. Then it has to re-encode the Phoenician glyphs into delicate subtle lightspeed ideas before it can do anything else. Blog The Specter Of Neuralese Sep 24, 2026
The science of reading these thoughts is a subfield of AI interpretability, which is still in its infancy. Blog Mysteries Of AI Generalization Sep 23, 2026
This trains the AIs to be focused on task success, which naturally risks including things like reward-hacking, cheating, and single-minded pursuit of stated goals at the expense of ethical injunctions. Blog The current balance of power in open models Sep 21, 2026
Since about April 2025, Chinese AI companies have been the clear leader in open-weight models. Blog The current balance of power in open models Sep 21, 2026
The simple metric showing this is Hugging Face Downloads, where China took the lead in July of 2025 primarily through the success of Alibaba’s Qwen models. About this quote 
Quotesby Glyph