Subvolts take Ilya Sutskever explains AI's move from scaling to research and AGI safety. Nov 25, 2025 - Developers
- Learn about new research directions beyond scaling and how to improve model generalization.
- Business
- Understand the evolving landscape of AI development and its implications for future AI capabilities.
- Everyday users
- Gain insight into the future of AI and discussions around ensuring its safe development.
Features a leading AI researcher discussing a fundamental shift in AI development strategy and AGI safety.
Ilya Sutskever — We're moving from the age of scaling to the age of research
Audio streams from Dwarkesh Patel. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Dwarkesh Patel interviews Ilya Sutskever on the Dwarkesh Podcast about the shift from scaling AI to research. They discuss SSI's strategy, pre-training issues, improving AI generalization, and AGI safety. The 96-minute episode explores moving beyond simply increasing model size to new avenues of AI development and ensuring beneficial AGI.
This episode discusses how AI development is changing. Instead of just making AI models bigger, researchers are focusing on new ways to make them smarter and safer. This matters because smarter, safer AI could help solve big problems.
This episode offers insights into the future direction of AI research from a prominent figure. It's for AI researchers, developers, and anyone interested in the long-term trajectory of artificial general intelligence. Sutskever's perspective on moving beyond scaling connects to current industry debates about model capabilities and safety, particularly relevant as labs like Google Gemini and OpenAI explore new architectures and training methods.
- HostDwarkesh Patel
- GuestIlya SutskeverSSI
- Guest: Ilya Sutskever
- Show: Dwarkesh Podcast
- Length: 96 minutes
- Main Topic: Shift from scaling to research in AI
- Discussed Topics: SSI's strategy, pre-training problems, AI generalization, AGI safety
Ilya & I discuss SSI’s strategy, the problems with pre-training, how to improve the generalization of AI models, and how to ensure AGI goes well.
Watch on YouTube; read the transcript.
Sponsors
* Gemini 3 is the first model I’ve used that can find connections I haven’t anticipated. I recently wrote a blog post on RL’s information efficiency, and Gemini 3 helped me think it all through. It also generated the relevant charts and ran toy ML experiments for me with zero bugs. Try Gemini 3 today at gemini.google
* Labelbox helped me create a tool to transcribe our episodes! I’ve struggled with transcription in the past because I don’t just want verbatim transcripts, I want transcripts reworded to read like essays. Labelbox helped me generate the exact data I needed for this. If you want to learn how Labelbox can help you (or if you want to try out the transcriber tool yourself), go to labelbox.com/dwarkesh
* Sardine is an AI risk management platform that brings together thousands of device, behavior, and identity signals to help you assess a user’s risk of fraud & abuse. Sardine also offers a suite of agents to automate investigations so that as fraudsters use AI to scale their attacks, you can use AI to scale your defenses. Learn more at sardine.ai/dwarkesh
To sponsor a future episode, visit dwarkesh.com/advertise.
Timestamps
(00:00:00) – Explaining model jaggedness
(00:09:39) - Emotions and value functions
(00:18:49) – What are we scaling?
(00:25:13) – Why humans generalize better than models
(00:35:45) – SSI’s plan to straight-shot superintelligence
(00:46:47) – SSI’s model will learn from deployment
(00:55:07) – How to think about powerful AGIs
(01:18:13) – “We are squarely an age of research company”
(01:20:23) – Self-play and multi-agent
(01:32:42) – Research taste
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com
- scaling
- Ilya Sutskever
- making AI models bigger and training them on more data.
- pre-training
- Ilya Sutskever
- the initial training phase of AI models on large datasets.
- generalization
- Ilya Sutskever
- an AI model's ability to perform well on new, unseen data.
- AGI
- Ilya Sutskever
- Artificial General Intelligence, AI with human-like cognitive abilities.
1 h 371 h 201 h 37AI researchers debate how close we are to recursive self-improvement
WithJohn Schulman, Beren Millidge, Charlie O’Neill
2 h 21Same guests, other shows
Questions people ask
- What is the main topic of the podcast episode?
- The episode discusses the shift in AI development from the age of scaling to the age of research, covering SSI's strategy, pre-training problems, and AGI safety.
- Who is interviewed on the Dwarkesh Podcast?
- The episode features an interview with Ilya Sutskever, discussing AI strategy and future research directions.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Dwarkesh Patel.