Ilya Sutskever to leave OpenAI

[zoogeny]

Interesting, both Karpathy and Sutskever are gone from OpenAI now. Looks like it is now the Sam Altman and Greg Brockman show.

I have to admit, of the four, Karpathy and Sutskever were the two I was most impressed with. I hope he goes on to do something great.

[nabla9]

Top 6 science guys are long gone. Open AI is run by marketing, business, software and productization people.

When the next wave of new deep learning innovations sweeps the world, Microsoft eats whats left of them. They make lots of money, but don't have future unless they replace what they lost.

[larodi]

Karpathy is still a mountain in the area of ML/AI, one of the few people worth following closely on Twitter/X.

[gdiamos]

I don’t think people give Dario enough credit

[davedx]

I love Karpathy. He's like a classical polymath, a scholar and a teacher.

[yu3zhou4]

Jakub Pachocki is still in OpenAI though

[albertzeyer]

Greg Brockman is a very good engineer. And that's maybe even more important in the current situation.

[ascorbic]

Jan Leike has said he's leaving too https://twitter.com/janleike/status/1790603862132596961

[Symmetry]

The scenario I have in my head is that they had to override the safety team's objections to ship their new models before Google IO happened.

[DalasNoin]

There goes the so called superalignment:

Ilya

Jan Leike

William Saunders

Leopold Aschenbrenner

All gone

[vintermann]

The guy with the "Bad universal priors and notions of optimality", which did to Hutter's MIRI program what Gödel did to Hilbert's program.

[snowbyte]

When walking around the U of Toronto, I often think that ~10 years ago Ilya was in a lab next to Alex trying to figure things out. I can't believe this new AI wave started there. Ilya, Karpathy, Jimmy Ba, and many more were at the right time when Hinton was there too.

[chollida1]

Oh man that was an amazing time at UoT. We also got GPU versions of btc mining from that group.

We also had Ethereum be born right around there as well around 2014. I remember the first Ethereum meetups around Queen and Spadina with Vitalik.

But to another posters point. Even though we had the father of deep learning Geoffrey Hinton and lumiaries like Ilya, and Vitalik, we didn't manage to get any real benefit from that.

[izend]

And none of them build AI companies in Toronto.

I’m Canadian and disappointed at how ineffective we are at building successful companies.

[ren_engineer]

seemed inevitable after that ouster attempt, probably just working out the details of the exit. But the day after their new features release announcement?

[ptero]

"Get next major feature to release and you can go as a friend" might have been part of an earlier agreement.

[twobitshifter]

I believe Omni was his work based on an interview he gave about end to end multimodal training being needed to move to the next level of understanding.

[gallerdude]

I would imagine he’d been thinking about it for a while, and maybe with all the buzz about him at the same time of the release, he was asked to decide.

[CooCooCaCha]

Could be a clever play. They sandwiched google io with news which has taken attention from Google. Plus they just had a big announcement so the negative news hits a little less hard.

[informal007]

People will pay relatively less attention on the leaving when announcement after new feature release than other any time.

[jakozaur]

Jakub Pachocki is amazing. He was in top 20 in Polish algorithm competition:

https://oi.edu.pl/contestants/Jakub%20Pachocki/

[rfoo]

Wait, TIL Jakub Pachocki == meret [1], never made the connection.

[1] https://codeforces.com/profile/meret

[treprinum]

The usual fate of idealistic people who build something great only to be discarded by management in a power struggle. How often did this repeat?

[wruza]

What do you mean how often, that is a foundation for the most successful economic model in humans. Some may not be discarded, but they will never get enough credit compared to a clueless head with a $1M smile talking to clueless heads with $1B wallets. We should thank god/nature that people who understand and do things exist in our species at all.

[sturza]

the people you need for the revolution are not the same you need after the revolution.

[darkerside]

Making sure generalized AI benefits everybody is the new Don't Be Evil

[unraveller]

"We want to put AI in your hands"

to keep??

NO! whatever gave you that idea, evil doer...

Open AI, as in, open your hands and beg for another hit of AI through thick rubber gloves and plexiglass.

[aaroninsf]

Why do people treat these technologists doing career moves, as if this was lineup changes in a major league sports teams?

Are these "first name" (ugh) "influencers" smart? Sure.

Smart is not that rare. These people are technologists like most of you, they aren't notably smarter, they just got lucky in their career direction and specialization. They aren't business geniuses.

They're just people filling roles.

Do changes in leadership affect a business? Sure? I guess? About 5% as much as you'd think from the tea-spilling gossip-rag chatter around AI people.

Enough already. Attend to the technology. Attend to the actual work. The number of you who are professionally impacted by these people changing paychecks is closer to zero than 50%.

[beacon294]

No, they are people defining companies which is a significantly less fungible placement and a self-defined role.

[TyrianPurple]

Meta's next for him? There's lots of money being poured into their AI division and there's lots of compute & being able to do any kind of research he might want.

[KaiserPro]

I doubt it, the internal politics of it are enough to drive most people crazy.

[Dowwie]

Does it matter that the people who dedicated the last decade to developing breakthrough work have left? It is a mistake to think that their luck streak will continue and their departure isn't a sign of decay at OpenAI. They may as well cash-in on their notoriety while it is of value. The odds are more in favor of other teams blazing new trails.

[fhd2]

Not to be a conspiracy theorist, but the phrase "So long, and thanks for everything" used in the tweet reminds me of "So long, and thanks for all the fish" from the dolphins in The Hitchhiker's Guide To The Galaxy. The background there is that dolphins are secretly more intelligent than humans, and are leaving Earth without them when its destruction is imminent (something the humans don't see coming).

I did once leave a company with a phrase just like that :P A few people there actually got the reference and congratulated me for the burn.

[zer00eyz]

I read Sam's Tweet and see "I fired him cause he voted against me"...

Im sorry but every time I see Sam speak, or read what he has to say all I can thing is "petulant man child".

> ... Ilya is easily one of the greatest minds of our generation

> ...Jakub is also easily one of the greatest minds of our generation

I'm not calling you a liar sam, but I just dont believe you.

[az226]

Trust was irrevocably broken. That’s why he is leaving.

[gnicholas]

I wonder how the proposed regulations to make noncompetes unenforceable affect moves like this. Or was he sufficiently high up that his existing noncompete would have survived?

[ed_mercer]

How good or bad is this for OpenAI?

[samspenc]

A few years ago? Probably catastrophic, he was Chief Scientist after all.

Now? Probably not too much, they have enough investment, and additionally talented people wanting to join. I mean, Andrej Karpathy also joined and left OpenAI twice and it didn't impact operations much.

I think OpenAI is now where Google was at or just before its IPO, a few key players leaving isn't going to impact them as much as it would have in its earlier founding days, and there is plenty of talent who are ready to jump in to fill the shoes of anyone who leaves.

[andsoitis]

According to Mr Altman’s tweet (https://twitter.com/sama/status/1790518031640347056) they had not just one but TWO of the greatest minds of this generation.

After this change they will have only one.

[spoonjim]

He is the smartest guy in AI but the sum of OpenAI’s talent is greater than his. But he could easily be the next great advancement in the field.

[az226]

Ilya hasn’t been working on core models for a while. He’s been focused on superalignment. That’s good for the world. Since OpenAI is leading/closest to AGI, it’s the best place to work on superalignment.

[lr4444lr]

Depends on how tight his non-compete is.

[jessenaser]

At least now we know GPT-5 has finished development and is now in training from this (I would hope that Iyla got to add all that he hoped to before leaving).

Ilya, thanks for all you have contributed within OpenAI!

[unraveller]

GPT-5ANDBAG more like it

He wouldn't have left if he could advance hoomanity further there, the guy has like a 800ms delay for each word and that does not make for a very good liar, perhaps a dutiful one.

[edmara]

From reporting GPT-5 finished pre-training while ago and was in the process of red-teaming.