Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI and Elon Musk

openai.com

Post reply on HN
Print view View original

851–860 of 1001 posts

Re: OpenAI and Elon Musk

#851

Post by olalonde »

samatman wrote: post #848
sigmoid10 wrote: post #313

Earlier quoted context omitted.

There is very little in terms of rigorous mathematics on the theoretical side of this. All we have are empirics, but everything we have seen so far points to the fact that more compute equals more capabilities. That's what they are referring to in the blog post. This is particularly true for the current generation of models, but if you look at the whole history of modern computing, the law roughly holds up over the l…

LLMs aren't improving at things they're unable to do at all. An example being reasoning.

LLMs can reason. You can verify this empirically by asking questions which require reasoning to e.g. GPT-4.

Re: OpenAI and Elon Musk

#852

Post by chaorace »

username332211 wrote: post #777
chaorace wrote: post #745

Earlier quoted context omitted.

It's actually a very clever structure! Please open the following image to follow along as I elaborate: https://pbs.twimg.com/media/F_PGOPOacAApU8e.jpg At the top level, there is the non-profit board of directors (i.e.: the ones Sam Altman had that big fight with). They are not beholden to any shareholders and are directly bound by the company charter: https://openai.com/charter The top-level nonprofit company owns ho…

And what was this structure supposed to achieve? At the top we have board of directors not accountable to anyone, except, as we recently discovered, to the possibility of a general rebellion from employees. That's not clever or innovative. That's just plain old oligarchy. Intrigue and infighting is a known feature of oligarchies ever since antiquity.

That's not entirely true. As a 501(c)(3) organization, they are bound to honor their founding bylaws on pain of having their tax-exempt status revoked w/ retroactive consequences. I won't comment on whether this is the fault of the bylaws or the IRS... but in the end I think we can agree that this was evidently not an effective enforcement mechanism.

As for the whole "cleverness" topic... it wasn't designed as an oligarchy, that's merely what it has devolved into. The saying "too clever by half" exists with good reason

Re: OpenAI and Elon Musk

#853

Post by daveguy »

chaorace wrote: post #849
daveguy wrote: post #804

Earlier quoted context omitted.

Except Altman has the political capital to have the entire board fired if they go against him, which makes the entire structure irrelevant. The power is where the technology is being developed -- at the bottom where they can threaten to walk out with plush jobs from the major shareholders at the bottom. The power is not where the figureheads sit at the top.

Right. That's the thing which I found to be so depressing. Just because I think that it was clever does not mean that I think it was successful. Out of curiousity, with the benefit of hindsight; what would you have tried doing differently to prevent such a coup?

That's a good question, and I agree it was a clever design. I don't know that there is a way to modify the org structure to prevent what happened. As much as I dislike them, a clear non-compete clause after the structure was in place might have helped, but I'm not sure that's even an option in CA. And having employees re-sign for noncompete would be fraught itself (better from the start). But this does seem like the most relevant application of non-compete. I'm not a lawyer and I'm sure they had top notch lawyers review the structure. If OpenAI played the non-compete card it wouldn't make retention easier if employees were willing to walk (they wouldn't exactly have trouble finding jobs anywhere). Do you know of anything that might have prevented it?

Re: OpenAI and Elon Musk

#854

Post by elif »

mort96 wrote: post #617
elif wrote: post #583

Earlier quoted context omitted.

Pandora's box is already open on that one.. and none of the model providers are really attempting to address that kind of issue. Same with impersonation, deepfakes, etc. We can never again know whether text, images, audio, or video are authentic on their own merit. The only hope we have there is private key cryptography. Luckily we already have the tools for this, NFT in the case of media and DKIM in the case of your…

Oh I definitely agree that there's no putting it back into the pandora's box. The technology is here to stay. I have no idea how you imagine "NFTs" will save us though. To me, that sounds like buzzword spam on your part.

An NFT is a way to attribute authorship in a mathematically guaranteed way.

If Taylor Swift signs ownership of a picture of her, you can know it is what she presents as real. If Elon musk signs a YouTube video of him offering crypto doubling giveaways, you can know he intended to represent the message as real. If the New York Times publishes an article signed with their key you can know it is meant to be published by the New York Times.

Re: OpenAI and Elon Musk

#855

Post by joquarky »

chaorace wrote: post #849
daveguy wrote: post #804

Earlier quoted context omitted.

Except Altman has the political capital to have the entire board fired if they go against him, which makes the entire structure irrelevant. The power is where the technology is being developed -- at the bottom where they can threaten to walk out with plush jobs from the major shareholders at the bottom. The power is not where the figureheads sit at the top.

Right. That's the thing which I found to be so depressing. Just because I think that it was clever does not mean that I think it was successful. Out of curiousity, with the benefit of hindsight; what would you have tried doing differently to prevent such a coup?

Ultimately, the people performing the actual work will always have a collective veto power.

Re: OpenAI and Elon Musk

#856

Post by chaorace »

daveguy wrote: post #853
chaorace wrote: post #849

Earlier quoted context omitted.

Right. That's the thing which I found to be so depressing. Just because I think that it was clever does not mean that I think it was successful. Out of curiousity, with the benefit of hindsight; what would you have tried doing differently to prevent such a coup?

That's a good question, and I agree it was a clever design. I don't know that there is a way to modify the org structure to prevent what happened. As much as I dislike them, a clear non-compete clause after the structure was in place might have helped, but I'm not sure that's even an option in CA. And having employees re-sign for noncompete would be fraught itself (better from the start). But this does seem like the…

Well, if you squint a little bit, this all looks kind of like a military coup. Through the cultivation of personal loyalty in his ranks, General Altman became able to ignore orders, cross the Rubicon, and subjugate the senate. It's an old but common story.

I point out this similarity because I suspect that the corporate solution to such "coups" will mirror the government solution: checks and balances. You build a system which functions by virtue of power conflicts rather than trying to prevent them. I won't pretend to know how such a thing could be implemented in practice, however.

Re: OpenAI and Elon Musk

#857

Post by renegade-otter »

elif wrote: post #56
Reubend wrote: post #13

Their evidence does a nice job of making Musk seem duplicitous, but it doesn't really refute any of his core assertions: they still have the appearance of abandoning their core mission to focus more on profits, even if they've elaborated a decent justification of why that's necessary to do. Or to put it more simply: here they explain why they had to betray their core mission. But they don't refute that they did betra…

>they may be right that open sourcing AGI would lead to too much danger. I think this part is proving itself to be an understandable but false perspective. The hazard we are experiencing with LLM right now is not how freely accessible and powerfully truthy it's content is, but it is precisely the controls upon it which are trying to be injected by the large model operators which are generating mistrust and a poor und…

ChatGPT is not about to run weapons systems. It's like throwing knives out the window and they complaining that knives are dangerous. Any automation is dangerous without due diligence.

Re: OpenAI and Elon Musk

#858

Post by xcv123 »

caeril wrote: post #778
habosa wrote: post #716

Regardless of who is right here, I think the enormous egos at play make OpenAI the last company I’d want to develop AGI. First Sam vs. the board, now Elon vs. everyone … it’s pretty clear this company is not in any way being run for the greater good. It’s all ego and money with some good science trapped underneath.

You're generally correct, but what really stings is Claude 3 Opus released right at the same time. It's superior to GPT-4 in pretty much every way I've tested. Center of gravity has shifted across a few streets to Anthropic seemingly overnight.

> It's superior to GPT-4 in pretty much every way I've tested.

They are just catching up to GPT-4 which was released a year ago (March 14, 2023). Meanwhile GPT-5 has been through a year of development.

Re: OpenAI and Elon Musk

#859

Post by pests »

gitaarik wrote: post #519
pests wrote: post #515

Earlier quoted context omitted.

Bad stuff? What bad stuff?

Well, you said it made them look bad. Or did you mean that they aren't doing anything bad but are only making it look bad for each other? Edit: sorry, I meant the parent said that. But I guess you understand it now. Edit 2: To be clear, the bad stuff: - OpenAI not really being "open" in any way. - Elon not really caring about that but just wanting to take revenge because they became successful without him.

That's not really "bad stuff" the public would care about IMO, just some internal drama.

Re: OpenAI and Elon Musk

#860

Post by samatman »

olalonde wrote: post #851
samatman wrote: post #848

Earlier quoted context omitted.

LLMs aren't improving at things they're unable to do at all. An example being reasoning.

LLMs can reason. You can verify this empirically by asking questions which require reasoning to e.g. GPT-4.

False. https://arxiv.org/abs/2308.03762
Post reply on HN