Scoop: theinformation.com
Paywalled link: https://www.theinformation.com/articles/three-senior-openai-...
Submitting information since paywalled links are not permitted.
I predict the board will be fired, and Sam and the team will return and try to contain the situation.
Do users care about that? I care about features stability and avoidance of shitification.
That's why I am usually preferring open models to depending on OpenAI's API. This drama has me curious about the outcome and if it leads to more openness from OpenAI, it may gain me back as a user.
Maybe not the individual users, but the enterprises/startups which builds around OpenAI.
I pay for ChatGPT, and I care.
What percentage of users, and how many in absolute numbers is a matter of debate, but this nonsense (and it is nonsense) is antithetical to building a strong trusting relationship with AI. At the very least it's as antithetical to their mission.
If we take a step back, the benchmark now is to be actually transparent. Radically transparent. Like when Elon purchased Twitter and aired all the dirty laundry in the Twitter Files transparent. The cowards at OpenAI hiding behind lawyers advising them of lawsuits are just that, cowards. Leaders stand by their principles in the darkest of times, regardless of whatever highfalutin excuses one could hide behind. It's pathetic and embarrassing. A lawsuit at a heavily funded tech startup at this level is not even a speeding ticket in the grand scheme of things.
95%+ of tech startup wisdom from the last decade is completely irrelevant now. We're living in a new era. The idea people will forget this in a month doesn't hold for AI. It holds for food delivery apps, not AI tech the public believes (right or wrong) might be an existential threat to their prosperity and economic future.
The degree of leadership buffoonery taking place at OpenAI is not acceptable and one must be genuinely stupid to defend it. Everyone involved should resign if they have any self-respect.
My prognostication is the market will express it's displeasure in the coming weeks and months, setting the tone for everyone else going forward. How the hell is anyone supposed to trust OpenAI after this?
The board and Ilya will all be gone within a month.
The board did become un-boardable in any future company, but they are not resigning.
In relation to other comments here. There is "coding" and there is "God's spark genius of algorithms" kind of work. This is what made the magic of OpenAI. Believe me, those guys were not "just coding". My bet is that it could be all about some research directions that were "shielded" by Sam.
I really don't buy that for a second. Most of OpenAI's value compared to any competitor comes from the money they spent hiring humans to trawl through training data.
They just hadn't -- and still haven't -- figured out how to commercialize it yet. I don't think they'll be the ones to crack that nut either. IMO they are too obsessed with "safety" to release something useful, and also can't reasonably deploy a service like ChatGPT at their scale because the costs are too high.
With OpenAI imploding this whole race just got a lot more interesting though...
As far as I can tell, all three of them are of Polish descent. For all we know they might have decided to resign together even if only one of them had a personal issue with OpenAI's vision. We will find out soon enough whether they will just found their own competing startup, based on OpenAI's "secret sauce" or not.
I'm not saying this will happen, but it seems to me like an incredibly silly move.
If he didn't manage to keep OpenAI consistent with it's founding principles and all interests aligned then wouldn't booting him be right? The name OpenAI had become a source of mockery. If Altman/Brockman take employees for a commercial venture, it just seems to prove their insincerity about the OpenAI mission.
Of course, not for the petty reasons that you list. Sama has comprehensively explained why the original OS model did not work, and so far the argument – it's very expensive – seems to align with a reality where every single semi-competitive available LLM (since they all pale in comparison to GPT-4 anyway) has been trained with a whole lot of corporate money. Meta side-chaining "open" models with their social media ad money is obviously not a comparative business, or any business. I get that the HN crowd + Elon are super salty about that, but it's just a bit silly.
No, Sam's failure as CEO is not having done what is necessary to align the right people in the company with the course he has decided on and losing control over that.
We barely understand how consciousness works, we should stop talking about "AGI". It is just empty, ridiculous techno-babble. Sorry for the harsh language, there's no nice way to drive home this point.
Should people discussing nuclear energy not talk about fusion?
And Microsoft are risk adverse enough that I think they do care about AI safety, even if only from a "what's best for the business" standpoint.
Tbh idc if we get AGI. There'll be a point in the future where we have AGI and the technology is accessible enough that anybody can create one. We need to stop this pointless bickering over this sort of stuff, because as usual, the larger problem is always going to be the human using the tool than the tool itself.
They may be geniuses, but AGI is an idea whose time has come: geniuses are no longer required to get us there.
The Singularity train has already left the station.
Inevitability.
Now humanity is just waiting for it to arrive at our stop
I think AGI is going to arrive via a different technology, many years in the future still.
LLMs will get to the point where they appear to be AGI, but only in the same way the latest 3D rendering technology can create images that appear to be real.
Not that I think AGI is possible or desirable in the first place, but that's a different discussion.
I can't imagine him doing that. He cares about getting well aligned AGI and profit motives fuck that up.
https://www.semanticscholar.org/author/J.-Pachocki/2713380?s...
As @eachro pointed out, Aleksander Madry is on leave from his MIT professorship. His publications:
If an AI said that, we'd be calling it "capability gain" and think it's a huge risk.
Maybe Ilya discovered something as head of AI safety research, something bad, and they had to act on it. From the outside it looks as if they are desperately trying to gain control. Maybe he got confirmation that LLMs are a little bit conscious, LOL. No, I am not making this up: https://twitter.com/ilyasut/status/1491554478243258368
With evidence, or is this the kind of pure speculation that media indulges in when they have no information and have to appear knowledgeable?
If they waited for the GPT5 pretraining to finish and then they minimized the cost of the loss of Altman and the engineers.
The whole secrecy, compartmentalization and urgency of their actions could only be explained by being against a wall. Otherwise if it was about ethics, future plans or whatever political it would happen at a slower pace.
Hope they involved their investors beforehand but I don't know if they had time, OpenAI probably still exists and evolves on other people's money. But what else could they do?
- he makes misleading statement to board
- board puts this in regulatory filing (e.g. SEC)
- board finds out this is a legally critical statement
- they _have_ to fire him in order to avoid becoming accomplices.
The reverse of the other Sam situation.
The OpenAI 501(c)3 already spun up a for-profit company in 2019 to do all the commercial work and take VC money.
Board had to act fast to fix it. And OpenAI changed enterprise pricing of API to be up front for cashflow related to that.
If you build out this way then when the next greatest LLM comes out you can plug that into your interface and switch the tasks it’s best at over.
AI tools will need a similar plugin like approach.
Hugely more interested in the open source models now, even if they are not as good at present. Because at least there is a near-100% guarantee that they will continue to have community support no matter what; the missing problem I suppose is GPUs to run them.
Don't know what to do. Is my investment into their API still worth it? It feels very unstable at this moment.
In all three languages I frequently use (Common Lisp, Python, and Racket) it is easy to switch between APIs. You can also use a library like LangChain to make switching easier.
For people building startups on OpenAI specific APIs, they can certainly protect themselves by using Azure as an intermediary. Microsoft is in the “stability business.”
They wouldn't have been able to do that even before Sam's dismissal
> Can they give reassurances about their products going into the future?
emotional comfort is not the thing you should be looking for mate.
The board of OpenAI should have been replaced by adults a long time ago.
These are huge losses. Pachocki led pre-training for GPT-4, and probably GPT-5. Brockman is the major engineer responsible for the efficiency improvements that enabled ChatGPT and GPT-4 to be even remotely cost-effective. That is a piece that is often overlooked, but OpenAI's advantage over the competition in compute efficiency is probably even larger than the model itself.
https://time.com/collection/time100-ai/6309033/greg-brockman...
It would be my life's dream to spend 80 hours per week coding without having to communicate with others... but no one is an island...
Did nobody read this damn article? It's 5 paragraphs long, take 1 minute.
As a part of this transition, Greg Brockman will be stepping down as chairman
of the board and will remain in his role at the company, reporting to the CEO.
[1] On one hand they serve Microsoft and developers, building digital AI infrastructure.
[2] On the other hand, they seem to try and want to build some monopoly and destroy as many startups (and companies) as possible.
In the last developer day, they did a half-ass job of both. GPTs suck. The OpenAI Assistants don't have enough documentation to be used and therefore, equally suck.
I really hope for the sake of the AI community (and economy) that [1] is the outcome. I really do not know how they could scale both. As an AI startup, I have a love-hate relationship with GPT and am eager to grow independent on them because how can I trust a company doing [2]?
Never change, hacker news!
To a person who is not an expert at prompting LLMs, ChatGPT is basically a shitty version of Bing Chat (aka Copilot). Especially the free version - it's an outdated model which cannot search the internet (or does it strictly worse than Bing Chat).
Why does OpenAI pay for access to a shitty version of Bing Chat?
There's only one possible reason for this: raising money at a very high valuation. They burned through hundreds of millions dollars to show VC that they have 100+ M users (and growing rapidly!) to raise at valuation ~$100B.
An exodus of staff might provoke a very quick release of a lot of code, pipelines, and training data by the remaining board before investors' lawyers have a chance to stop it.
Sam also talked about the dangers of AI. It’s likely that he did so to encourage a regulatory moat around OpenAI.
If there was some real misconduct by Altman, others won't be resigning with him, would they?
Advice: you can't win over a narrative, which is what Sam has become. People and resources will come to him, by themselves.
If I was OpenAI I'd want to quit today not because I want to follow Sam, but because the same bullshit that people left Google Brain et al. for has managed to catch up with them at OpenAI. It's a shame honestly, it was so exciting to see a company finally free itself of the shackles of navel gazing and just build things, but it seems like that's over today.
The narrative I am referring to is a simple one: take back what should be his. A.k.a, revenge. That is for sure a strong word.
People live on stories (collective imagination, if fancier), and what they like most is a wronged prince/princess took back his/her crown. It is the same with Taylor Swift rerecording her albums. The story potential will feed itself, until it is fully realized. The OpenAI board had committed a historical misstep, but maybe it is indeed what it is designed for: they hold no stakes in the business, so their view can't be held accountable in a business microscope. But money will really dislike it.
the dataset is the crucial bit of openAI. that takes a lot of time and money to make. So its perfectly possible for openAI to carry on innovating without these people.
But equally, it could turn to shit.
However, Sam isn't jesus, so its not like he's going to magically make another successful company.
I bet he'll train models on copious amounts of synthetic data made with GPT-4. There are lots of datasets in the open. That makes catching up easier.
No public facing model can be protected from data exfiltration and distillation. All deployed skills leak, your competition will replicate with less effort. And they only need to leak once and every subsequent model can inherit the skill. I think the first movers paid a high price for being first, and will quickly see their advantage erode. Latecomers will catch up and find AI easier to work with. The difference is made by the great fine-tuning datasets that are in the open, a growing lake of distilled abilities.
Another latecomer advantage is benefiting from significant innovation in the engineering part: flash attention, quantization, continuous batching, KV caching, LoRA, and more.
The new AI era will be more equalitarian. Catching up is much easier than discovering, and we can run AI privately, unlike search engines and social networks. You can't exploit SOTA advantage at scale. Being first is a fleeting advantage, the moment you go in the open everyone replicates.
Maybe one reason this is happening is because AI skills are very composable. Any addition to the skill repertoire already fits with other skills. This makes open sourcing skills very attractive. Of course, the datasets are what is being open sourced.
But I think one thing is certain, he WILL create another AI company. It seems very unlikely he would quit the business.
Madry is actually spelled Mądry, which translates to "Wise".
Anyway, I was providing context. The way I see it, since they speak the same language natively, they might have had a frank conversation about this at some point and decided to resign in concert.
Im not claiming to know more than everyone else, but Sam was IMO just a face. Greg is backend engineer - that's less important than actual research.
For profit will always win non-profit because It’s run by people - once they see money , they want more not less.
This was meant to happen when it changed its structure to cap profits at 100x . If Microsoft invests $10 billion openAI Can make $1 trillion in profits before giving a dime to its non profit. If it makes $1 trillion it can just ask someone to invest $1 billion and wait until it makes another $100 billion .
Sam, Brockman and others who resigned will start a new for-profit company. Sam will use his influence to get funding and contracts. They know the tech and with know how of everything, build better model. And OpenAI will be forced to give up and work with them as it’s part of their founding policy.
"Goddamn, how dare they invent the bigger cannons!?" - Romans, 1453, in Constantinople probably. (One incident where I can use my exempt powers).
Have you used gpt4? it's reasoning capabilities match human ability. The more you think about it, the more scary the reality becomes. GPT 4 can reason through any mental exercise as well as a human. The rest of the work to make it autonomous is simple in comparison.
Satya has been humiliated and will be furious.
Microsoft has a minority stake in the for-profit subsidiary that is wholly controlled by the 501(c)(3). All investors (and employees) in the for-profit have to agree to the operating agreement that specifies the for-profit is not actually obligated to actually make a profit and that it is all secondary the charter the non-profit operates under.
https://openai.com/our-structure https://openai.com/charter
There is not a higher power than the board of the non-profit.
I think people seriously underestimate how hard it is to get the GPU compute/etc that is necessary to be useful here. Lead time would be years, easily, even if you had the money. NVidia can't change this for you even if they liked you - they literally can't build chips fast enough.
Depending on the exact agreement, Microsoft may have just given them credits/free use, and the part where they make sure the resources are actually available is just good faith that may no longer exist.
That's one example.
Even beyond that, you are assuming they can only do super-direct things, but it turns out to be fairly easy to make things very uncomfortable for people/companies indirectly.
Only time employees leaving is a bad sign had been when the company had been in zombie mode too long(few years) and haven’t delivered anything yet or still in alpha/beta stage with crappy deliveries no one is using or has no significant uptake.
Also, if there are financial difficulties, then a lot of people leave in droves and some actually come back after finances get better.
it's like you wrote this comment yesterday, and not in the context of what has happened
My guess is they know something we don't Or they assume sama being fired means the trajectory of OpenAI as a company has changed or is likely to change significantly?
Would it really have been so hard to find out Sam Altman just got fired without notice?
Presumably it would require enormous startup capital?
I think they literally could get dump trucks full of cash.
I see no reason why they would of course.
Still as far as I understand and I fully admit that Is not a lot, OpenAI is running on a lot of MS cloud, so they would (probably ) ot be able to offer OpenAI² enough compuete in the near future.
I doubt anyone could, but if it became a need I am certain every at scale provider would be doing their upmost to win the business.
I’m not making any judgement about whether they made the correct decision. I’m just stating that everyone keeps talking about this as if it were a normal company structure and it absolutely is not.
So here's my theory which might sound crazy. Sam planned to open a new AI company and taking away openAI's top talents to his company. And breaking up openAI into non profit and his for profit company.
Sam's first tweet after all this has, just hours after this article:
> will have more to say about what’s next later.
So either he knew that he was about to be fired or at least was prepared.
Also based on the wording of the press release, Sam did something that the board absolutely hated. Because most of the time even if he did something illegal it doesn't make sense to risk defamation by accusing him publically. Also based on his video of yesterday at the APEC summit, he repeated the similar lines few times:
> I am super excited. I can't imagine anything more exciting to work on.
So here if we assume he knew he was about to get fired, the conclusion is clear.
openai was that backed by a nonprofit structure. and it still caused sam to be michael dell'ed/steve job'ed.
seems like the issue was with having a board/didnt have a majority on it. Zuckerberg having 53% voting power on the board is probably the greatest thing he managed. Anything sam does from now should follow the same.
/disclaimer - i have no idea how voting shares work.
First sentence on https://openai.com/about
(it's about just one letter - a vs ą)
If that's the way he wants to be known, it's up to him.
I've started a directory of GPTs on gipety.com so personally the longer it takes for the official store to be launched the better ;)
This rule should be dropped. It's the reason that HN is dominated by low quality outlets. It's also in tension with another rule, which is that the original source of the article should be posted - the original source is usually an outlet that has the resources to do original research, thanks to a paywall.
They are? Here's my submission of your link.
https://news.ycombinator.com/item?id=38320350
Maybe @dang can put it under your name?
How much more cost effective to just fool half the board of a nonprofit into taking unnecessarily aggressive action.
I guess he was trying to seem cool and approachable, but to me it just reads as unprofessional. I find it hard to take a piece of writing seriously when it’s formatted that way.
It might indicate intelligence and people who are extra busy, thus not wasting their time, but depending on the position and circumstances a nicely written text is essential.
I guess that’s what happens if you’re working 60-100 hours per week coding ChatGPT, and you don’t have time to waste. Sometimes I feel sorry for these kind of people and other times I envy them.
imagine being someone that is just being quick, and everyone around you is inferring some motive like 'being cool', when really he isn't thinking of your perceptions at all.
Ironically some of us haven't gotten used to this yet either.
But it does look very weird. the full stop at the end of a capitalless sentence followed by an ellipsis is what sticks out to me, seems very tryhard.
The fact that it’s all well connected wealthy people is kind of the point, the board is there (among others) to bring advice and experience
OpenAI non-profit --owns---> Holding Company for Open AI non-profit
That makes no sense. How does a non-profit "own" a holding company, that own the non-profit?
He was the one who partnered with Microsoft and turned it from a non-profit to a for-profit company.
It's completely illogical to say that just because an AI company isn't open source it can't strive to also be "safe".
The truth is there isn't any strong evidence one way or another, only your existing worldviews.
Partnering with Microsoft and closing access both have profit-driven and ideology driven explanations, and we don't have strong evidence it's one and not the other.
If you're that much of a narcissist that you'd risk your health to look better, I need to take everything else you do and say with a big grain of salt.
In the world where Sam Altman leads OpenAI to market dominance and eventual acquisition by Microsoft for $400B or whatever, he obviously would represent an important part of what Microsoft would be buying and would be compensated accordingly.
That's a good sign imo.
It's a bubble if the valuation is inflated beyond a level compared to a reasonable expected future value. Usefulness isn't part of that. The important bit is 'reasonable', which is also the subjective bit.
We've already trained it on all the data there is, it's not going to get "smarter" and it'll always lack true subjective understanding, so the overhype has been real, indeed to bubble levels as per OP.
https://en.wikipedia.org/wiki/Embrace,_extend,_and_extinguis...
Literally people thought they were saints because of VSCode FFS.
In other news, Tesla FSD has been rebranded Saneer-Weeksbooth Autobot. So be careful out there folks.