Ilya Sutskever on AI
12 quotes from Ilya Sutskever on the wall, 2017 to 2025, newest first. Each one is verbatim and links to where it was said. Sources include Business Insider, dwarkesh.com, garymarcus.substack.com.
- 2025
These models somehow just generalize dramatically worse than people
Ilya Sutskever, OpenAI co-founder and deep learning researcher, in a just-released interview discussed on Gary Marcus's Substack, November 2025, on the limits of scaling large language models.
garymarcus.substack.comA trillion dollars is a terrible thing to waste
AI is going to be both extremely unpredictable and unimaginable
Ilya Sutskever, OpenAI co-founder, speaking in a video interview with The Open University of Israel, June 2025.
The Hans IndiaIlya Sutskever Warns of an Unpredictable AI Future: "It's Going to Be Unimaginable"
- 2024
Name of Ilya Sutskever's company Safe Superintelligence Inc., founded June 2024, as set out in its founding statement on ssi.inc.
- 2023
The beatings will continue until morale improves
Ilya Sutskever posted this line on X in December 2023, shortly after OpenAI's leadership turmoil over Sam Altman's brief ouster and reinstatement; he deleted it the next day.
I deeply regret my participation in the board's actions
Ilya Sutskever, OpenAI co-founder, chief scientist and board member, in a post on X on November 20, 2023, after voting to fire Sam Altman as OpenAI CEO. He later left the board when Altman was reinstated.
Oh my god, this computer seems to understand
Ilya Sutskever, OpenAI cofounder and chief scientist, describing the reaction of first-time ChatGPT users in an interview with MIT Technology Review, as reported by Business Insider, October 2023.
Many people will choose to become part AI
Ilya Sutskever, OpenAI's chief scientist, speaking to MIT Technology Review in October 2023 about how humans might keep up with superintelligent AI. He said he was unsure whether he would choose it himself.
It just does not make sense to open-source
Ilya Sutskever, OpenAI co-founder, told The Verge in 2023 that open-sourcing advanced AI models is a bad idea once AGI becomes possible.
VoxWhy Meta is giving away its extremely powerful AI model - vox.com
Predicting the next token well means that you understand the underlying reality
Ilya Sutskever, OpenAI co-founder and chief scientist, in a March 2023 Dwarkesh Podcast interview (introduced as The Lunar Society), explaining what next-token prediction implies about a model's understanding of the world.
I challenge the claim that next-token prediction cannot surpass human performance
Ilya Sutskever, OpenAI co-founder and chief scientist, in a March 2023 Dwarkesh Podcast interview with Dwarkesh Patel, introduced on air as The Lunar Society, discussing whether next-token prediction can exceed human-level intelligence.
- 2022
It may be that today's large neural networks are slightly conscious
Ilya Sutskever on X, February 2022.
- 2017
Sutskever to Musk and Altman, 20 Sep 2017: "The goal of OpenAI is to make the future good and to avoid an AGI dictatorship."
Mentioned in other quotes 6
Quotes by others that refer to Ilya Sutskever
- 2026
Chant at OpenAI, reportedly led by chief scientist Ilya Sutskever (The Atlantic, November 2023).
- 2025
Now we really know what Ilya saw
Nathan Lambert, in Interconnects analysis of GPT-4.5's release, February 2025, referring to reports that Ilya Sutskever had foreseen limits to pretraining scaling.
- 2023
Superintelligence—AI vastly smarter than humans—could be developed within the next ten years
OpenAI's blog post accompanying its December 2023 'weak-to-strong generalization' paper on superalignment, the first from Ilya Sutskever and Jan Leike's Superalignment team, predicting the arrival of superintelligent AI.
GizmodoIlya Sutskever’s Team at OpenAI Built Tools to Control a Superhuman AI
I harbor zero ill will towards him
OpenAI CEO Sam Altman, in a message to OpenAI staff on November 29, 2023, after his reinstatement, speaking about co-founder Ilya Sutskever after the board upheaval and his departure from the board.
Could lead to the disempowerment of humanity or even human extinction
OpenAI statement on the dangers of superintelligence, announcing a new alignment research team led by co-founder and chief scientist Ilya Sutskever, with 20 percent of its processing capacity, July 2023.
theregisterOpenAI says it is dedicating compute to stopping 'rogue' AI - The Register
- 2013
Nonintuitive characteristics and intrinsic blind spots
Christian Szegedy of Google with coauthors including Ilya Sutskever and Ian Goodfellow, in the December 2013 paper "Intriguing properties of neural networks" (ICLR 2014) that first reported adversarial examples in deep networks.