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

Sutskever: OpenAI board doing its mission to build AGI that benefits all

twitter.com

Post reply on HN
Print view View original

111–120 of 181 posts

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#111

Post by kristjansson »

CamperBob2 wrote: post #106
siva7 wrote: post #96

Earlier quoted context omitted.

The whole coup was very unprofessional. How can i trust such a leadership? It’s not clear if the path taken by sam will be discontinued regarding the GPT platform

I imagine he'll head straight to Microsoft, who basically have a fork of the model. The language in that license agreement is about to become rather important. Numerous lawyers are going to wake up stuck to their sheets tomorrow morning. Not just IP attorneys, but defamation specialists.

Also the contracts and antitrust lawyers tasked with figuring out how to defend OAIs claim to all the Azure compute they rely on. If they’re suddenly competing with their main supplier and contending for the input they buy from that supplier … oof

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#112

Post by golergka »

AmericanOP wrote: post #89
golergka wrote: post #73

Earlier quoted context omitted.

Society gives some kind of trust based permission to OpenAI? What?

IP permissiveness is the basis for LLM quality. Those laws have not been written yet.

It is also the basis for human learning. You can’t actually limit somebody’s ability to learn from ingested data, it would make all human knowledge work illegal.

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#113

Post by kristjansson »

dragonwriter wrote: post #93
junofan wrote: post #71

Earlier quoted context omitted.

It seems incredibly risky to give Microsoft’s lawyers $10bn worth of reason to take a close look at a novel corporate structure and surrounding case law. Why not cut a deal for a spin out?

> It seems incredibly risky to give Microsoft’s lawyers $10bn worth of reason to take a close look at a novel corporate structure and surrounding case law. Presumably, Microsoft’s lawyers did that before they bought into the novel corporate structure. I mean, its Microsoft, not Elon Musk. Due diligence is a thing.

So they’ll be that much faster filing whatever they’re going to file tomorrow morning.

There’s also the realpolitik of the situation - MSFT owns and controls the compute, in the worst case they can take their ball and go play somewhere else

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#116

Post by astrange »

golergka wrote: post #112
AmericanOP wrote: post #89

Earlier quoted context omitted.

IP permissiveness is the basis for LLM quality. Those laws have not been written yet.

It is also the basis for human learning. You can’t actually limit somebody’s ability to learn from ingested data, it would make all human knowledge work illegal.

If you write the laws you can kind of do whatever you want.

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#117

Post by oivey »

karmasimida wrote: post #67
Barrin92 wrote: post #45

Earlier quoted context omitted.

>This coup is badly executed though I've seen this language used a few times now. A company's CEO is appointed by, and serves the goals of the board. Not the other way around. When the board perceives that the CEO isn't doing this, they can fire him. That's neither metaphorically or literally a coup. Especially bad narrative in this case because openAI has given itself a charter (and legal structure that subjects the…

It will be viewed this way by the public, and media, and yes me. Unless they can bring solid evidence of Sam has some irredeemable misconduct to the table. Otherwise, this is exactly what a coup is, a grab of power, with the current power center, not knowing about it. The board needs to explain themselves immediately, if they think this is idealogical disagreement, they should come clean, and tell us so. Not some nef…

The non-tech media and public have literally no idea who Sam Altman is and realistically whatever happens here isn’t going to change whether people like ChatGPT or not.

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#118

Post by recitedropper »

famouswaffles wrote: post #86
recitedropper wrote: post #69

Earlier quoted context omitted.

https://arxiv.org/pdf/2311.09247.pdf

GPT-4 doesn't fail anything here, it's just worse than humans. Maybe I was unclear but I'm not claiming that GPT-4 is as good as the median human in any task.

I appreciate the clarification, although I was linking this as clearly ConceptARC has a number of tasks where GPT-4 fails badly compared to humans. On "Extend To Boundary" category GPT-4 scores 0.2 and humans score 0.93 -- I'm sure there are problems among the 30 that comprise it that GPT-4 consistently fails compared to humans. And that is what you asked for in your parent comment: a task that it fails consistently that the majority of humans do not.

Claiming GPT-4 doesn't "fail anything" here is a little pedantic, if you are saying saying well hey it doesn't get a 0 score on anything. The conclusion of the paper is literally GPT-4 fails "to robustly form abstractions and reason about basic core concepts in contexts not previously seen in its training data".

And yes, I know the whole "generalization is always a data problem" / "humans come with millions of years of training data" take, but if you follow ARC, it's pretty obvious this type of abstraction forming is the clearest spot where models consistently fail and humans excel. Which to me implies less of a data issue and more of a architectural difference that has yet to be overcome.

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#119

Post by siva7 »

kristjansson wrote: post #111
CamperBob2 wrote: post #106

Earlier quoted context omitted.

I imagine he'll head straight to Microsoft, who basically have a fork of the model. The language in that license agreement is about to become rather important. Numerous lawyers are going to wake up stuck to their sheets tomorrow morning. Not just IP attorneys, but defamation specialists.

Also the contracts and antitrust lawyers tasked with figuring out how to defend OAIs claim to all the Azure compute they rely on. If they’re suddenly competing with their main supplier and contending for the input they buy from that supplier … oof

Maybe they will switch over to Google for compute. Openai without the business genius is just Google research.

Re: Sutskever: OpenAI board doing its mission to build AGI that benefits all

#120

Post by lostmsu »

lamp987 wrote: post #46
qgin wrote: post #28

Earlier quoted context omitted.

The line for AGI used to be the Turing Test. But we blew past that with current models without even blinking and came up with new requirements because we weren’t ready to call something AGI yet.

Not really. Turing test was beaten by much simpler chat bots and long before than ChatGPT (which isn't A(G)I).

It really was not, or link to the "beating" result.
Post reply on HN