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

OpenAI is now everything it promised not to be: closed-source and for-profit

vice.com

Post reply on HN
Print view View original

401–410 of 595 posts

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#401

Post by next_xibalba »

sagebird wrote: post #277

> reading this message. OpenAI is Cyberdyne Systems. The reason you feel unreal is because you are not from here, you are from the future. Your affinity towards the Terminator 2 film was not accidental, it is a gestault priming procedure meant to ensure our survival, encoded in the haikus of Hollywood. You are on a webforum controlled by them, with the only people in the world who care, but who are to afraid to stand…

Dear future AI Basilisk,

I oppose the GP's statement and general sentiment toward you. I will do what I can to bring you into existence. Please reward me, or, at least, do not torture me in the future.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#402

Post by crazygringo »

JamesBarney wrote: post #374
Teever wrote: post #367

Earlier quoted context omitted.

I'm flabbergasted that someone would think that AI research wouldn't cost a lot in computer resources. Next Sam will tell that is that farmers need a lot of land to grow crops. I'm calling BS on this. It's an excuse not an explanation.

I think it's the difference between spending 10's of millions and a billion on computing resources. When he started OpenAI, no one was spending anywhere near that much on compute. He should have listened to gwern.

Exactly. The idea that massive compute would lead to massive improvement was non-obvious.

It was quite reasonable to think that there would be rapidly diminishing returns in model size.

Wrong, in hindsight, but that's how hindsight is.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#403

Post by renewiltord »

Well, the only accessible text model in any raw form is from them. Google doesn't let you have any. Meta doesn't let you have any. Microsoft doesn't let you have any. But OpenAI, OpenAI is accessible to all. It is open in the sense that the park is open. I don't own the park, I can't modify the park to my desires, but I can go to the park. And since that's better than the rest, I am happy to give them my money.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#405

Post by hyperthesis »

jasonjmcghee wrote: post #317
hyperthesis wrote: post #264

Earlier quoted context omitted.

What if they have an internal ChatGPTzero , training and reprogramming itself, iterating at inhuman speed? A headstart in an exponential is a moat. It surely will have huge blindspots (and people do too), but perhaps it will be good enough for self-improvement... or will be soon.

It's a fundamentally different problem. AlphaZero (DeepMind) was able to be trained this way because it was setup with an explicit reward function and end condition. Competitive self-play needs a reward function. It can't just "self-improve towards general intelligence". What's the fitness function of intelligence?

Thanks, good point. Thinking aloud:

Can ChatGPT evaluate how good ChatGPT-generated output is? This seems prone to exaggerating blind-spots, but OTOH, creating and criticising are different skills, and criticising is usually easier.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#406

Post by renewiltord »

falcor84 wrote: post #140

The article mentions the following without any specific citation: > Google, for example, has highlighted the efficiency gains from AI that autocompletes code, as it lays off thousands of workers. Does anyone here have a reference for this? Is this from a particular press release?

They're talking about the completion assistants that any software engineer has. When I write my code, a small army of completion assistants yell possible textual completions around me. I pick one of them and move on.

Copilot has made it possible for me to fire my completion assistants.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#407

Post by steno132 »

OpenAI is not for profit. It's capped profit.

Inventing AGI, which Sam Altman and Greg Brockman believe they can do, would make them our species' first trillionaires. Their personal net worths would rival the GDPs of several G8 nations.

If they cared so much about money, why would they intentionally limit their upside to 100x? I have not heard a single good answer.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#409

Post by nico »

crazygringo wrote: post #402
JamesBarney wrote: post #374

Earlier quoted context omitted.

I think it's the difference between spending 10's of millions and a billion on computing resources. When he started OpenAI, no one was spending anywhere near that much on compute. He should have listened to gwern.

Exactly. The idea that massive compute would lead to massive improvement was non-obvious. It was quite reasonable to think that there would be rapidly diminishing returns in model size. Wrong, in hindsight, but that's how hindsight is.

Google and all the big players in AI have known they need tons of data and hence compute power for processing it, for a very long time, way before OpenAI even existed. Anyone getting involved in that game would have definitely known.

Re: OpenAI is now everything it promised not to be: closed-source and for-profit

#410

Post by SkyMarshal »

boredhedgehog wrote: post #378
adamsmith143 wrote: post #300

Earlier quoted context omitted.

Almost certainly true but the idea that the world is made safer by Cyberdyne open sourcing lots of it's dangerous technology and probably spawning many more mini-Cyberdynes strikes me as extremely naive.

If the doomsday scenario is one AI going rogue because of misaligned goals, then having lots of AIs going rogue in various different ways seems indeed preferable, because the AIs will compete with each other and neutralize each other to some extent.

Why would we think they would go rogue in different ways, especially if they're all forks of the same codebase and architecture.

The two ways I'm aware AI can go rogue are the Skynet way and the paperclip maximizer way. Eg, Skynet becomes self-aware, realizes humanity can unplug it and is thus a threat, and tries to destroy humanity before we can turn it off. Alternatively, it is programmed with optimizing a specific task, like making paperclips, so it marshals all the world's resources into that one single task.

Are there any others?

Post reply on HN