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

Emad Mostaque resigned as CEO of Stability AI

stability.ai

Post reply on HN
Print view View original

211–220 of 429 posts

Re: Emad Mostaque resigned as CEO of Stability AI

#211

Post by sigmoid10 »

HlessClaudesman wrote: post #197
famouswaffles wrote: post #45

Earlier quoted context omitted.

Midjourney is the most popular discord channel by far with 19.5M+ members, $200M in revenue in 2023 with 0 external investments and only 40 employees. The problem has nothing to do with commercializing image gen AI and all to do with Emad/Stability having seemingly 0 sensible business plans. Seriously this seemed to be the plan: Step 1: Release SD for free Step 2: ??? Step 3: Profit The vast majority of users couldn'…

Heres a Stable Diffusion buisness idea: sign up all the celebrities and artists who are cool with AI, and provide end users / fans with an AI image generation interface, trained on their exclusive likenesses / artwork (loras). You know, the old tried and true licensed merchandise model. Everybody gets paid.

You can already do that with reference images and even for inpainting. No training required. Also no need to pay actors outrageous sums to use their likeness in perpetuity as long as you do business. The licensing still tricky anyways, because even if the face is approved and certified, the entire body and surroundings would also have to be. Otherwise you basically re-invented the celebrity deepfake porn movement. I don't see any A-lister signing up for that.

Re: Emad Mostaque resigned as CEO of Stability AI

#212

Post by AnthonyMouse »

fauigerzigerk wrote: post #199
AnthonyMouse wrote: post #144

Earlier quoted context omitted.

You can get registered DDR4 for ~$1/GB. A trillion parameter model in FP16 would need ~2TB. Servers that support that much are actually cheap (~$200), the main cost would be the ~$2000 in memory itself. That is going to be dog slow but you can certainly do it if you want to and it doesn't cost $50,000.

How slow? Depending on the task I fear it could be too slow to be useful. I believe there is some research on how to distribute large models across multiple GPUs, which could make the cost less lumpy.

You can get a decent approximation for LLM performance in tokens/second by dividing the model size in GB by the system's memory bandwidth. That's assuming it's well-optimized and memory rather than compute bound, but those are often both true or pretty close.

And "depending on the task" is the point. There are systems that would be uselessly slow for real-time interaction but if your concern is to have it process confidential data you don't want to upload to a third party you can just let it run and come back whenever it finishes. And releasing the model allows people to do the latter even if machines necessary to do the former are still prohibitively expensive.

Also, hardware gets cheaper over time and it's useful to have the model out there so it's well-optimized and stable by the time fast hardware becomes affordable instead of waiting for the hardware and only then getting to work on the code.

Re: Emad Mostaque resigned as CEO of Stability AI

#213

Post by janalsncm »

stale2002 wrote: post #182
janalsncm wrote: post #176

Earlier quoted context omitted.

Not being e/acc doesn’t make one a doomer. It means being someone with a healthy relationship with technological progress, who hasn’t given up on the prospect of useful regulation where it’s helpful. The US is great but the best AI researchers aren’t gonna want to live here if it becomes a hyper libertarian hellscape. They want to raise a family without their kids being exploited by technologies in ways that e/accs t…

> if it becomes a hyper libertarian hellscape. Releasing cool open source AI tech doesn't turn the world into a libertarian hellscape. You are taking the memes way too seriously. Mostly people just joke around on twitter while are also building tech startups. e/acc isn't overthrowing the government.

You may want to read Marc Andreessen’s manifesto. E/acc isn’t just about “releasing cool AI tech”.

It means unrestricted technological progress. Unrestricted, including from annoying things like consumer protection, environmental concerns, or even national security.

If e/acc was just about making cool open source stuff and posting memes on the Internet, you wouldn’t need a new term for it, that’s what people have been doing for the past 30 years.

Re: Emad Mostaque resigned as CEO of Stability AI

#214

Post by astrange »

immibis wrote: post #188
astrange wrote: post #78

Earlier quoted context omitted.

No, that's the sign they're being sensible, since the thing you want is illegal in most countries.

I looked this up and apparently the controversy is that Stable Diffusion was trained on child porn? How can the model itself not be considered objectionable material then? Does the law not apply some kind of transitive rule to this? And don't they want to arrest someone for having the child porn to train it on?

There's two controversies that are the opposite of each other.

1. Some people are mad that Stable Diffusion might be trained on CSAM because the original list of internet images they started with turned out to link to some. (LAION doesn't actually contain images, just links to them.)

This one isn't true, because they removed NSFW content before training.

2. Some other people are mad that they removed NSFW content because they think it's censorship.

That actually isn't the legal issue I meant though. It's not that they trained on it, it's that it contains adult material at all (and can be shown to children easily), and that it can be used to generate simulated CSAM, which some but not all countries are just as unhappy about.

Re: Emad Mostaque resigned as CEO of Stability AI

#215

Post by sigmoid10 »

mirekrusin wrote: post #165
ctrw wrote: post #32

Earlier quoted context omitted.

Think seti at home. Instead of wasting all the compute on bitcoin we pretrain fully open models which can run on people's hardware. A 120b ternary model is the most interesting thing in the world. No one can train one now because you need a billion dollar super computer.

SETI made sense because there is a lot of data where you download chunk and do expensive computation and return thin result. Model training is unlike that. It's large state that is constantly updated and updates require full, up to date state. This means you cannot distribute it efficiently over slow network with many smaller workers. That's why NVIDIA is providing scalable clusters with specialized connectivity so t…

That's wrong. What you described is data parallelism and it would indeed be very tricky to e.g. sync gradients across machines. But this is not the only method of training neural nets (transformers or any other kind) in parallel. If we'd like to train, say, a human brain complexity level model with 10^15 parameters, we'd need a model parallelism approach anyways. It introduces a bit of complexity since you need to make sure that each distributed part of the model can run individually with roughly the same amount of compute, but you no longer need to worry about syncing anything (or have the entire state of anything on one machine). The real questions is if you can find enough people to run this who will never be able to run it themselves in the end, because inference alone will still require a supercluster. If you have access to that, you might as well train something on it today.

Re: Emad Mostaque resigned as CEO of Stability AI

#217

Post by johnnyanmac »

raincole wrote: post #163
bamboozled wrote: post #44

Earlier quoted context omitted.

Tens of millions of freeloaders too.

Every single Ad blocker users is a freeloader to Google/Youtube. Didn't stop them from being extremely successful.

studios are the target here, not consumers. Pareto principle applies when you need more than a passive user. 20% or less of the serious studios (which is already a minority) will end up providing 80% of the value of any given AI solution.

Re: Emad Mostaque resigned as CEO of Stability AI

#218

Post by stale2002 »

janalsncm wrote: post #213
stale2002 wrote: post #182

Earlier quoted context omitted.

> if it becomes a hyper libertarian hellscape. Releasing cool open source AI tech doesn't turn the world into a libertarian hellscape. You are taking the memes way too seriously. Mostly people just joke around on twitter while are also building tech startups. e/acc isn't overthrowing the government.

You may want to read Marc Andreessen’s manifesto. E/acc isn’t just about “releasing cool AI tech”. It means unrestricted technological progress. Unrestricted, including from annoying things like consumer protection, environmental concerns, or even national security. If e/acc was just about making cool open source stuff and posting memes on the Internet, you wouldn’t need a new term for it, that’s what people have bee…

Ok, thats nice and all. But the actual results of all of this is memes and people making cool startups.

Regardless of whether a couple people who are taking their own jokes too seriously truly believe that they are going to, I don't know, create magic AGI, the fact remains that the actual measurable results of all this is only:

1: funny memes

2: cool AI startups

Anything other than that is made up stuff in either your head, or the heads of people who just want to pump up the valuation of their startups.

> Marc Andreessen’s manifesto

Yes, I'm sure he says a lot of things that will convince people to invest in companies that he also invests in. It is an effective marketing tactic. I'm sure he has convinced other VCs to invest in his companies because of these marketing slogans.

But regardless of what a couple people say to hype up their startup investments, that is unrelated to the actual real world outcomes of all of this.

> you wouldn’t need a new term for it

The fact that me and you are talking about it, actually proves that yes some marketing terms both make a difference and also don't result in, I don't know, the government being overthrown and replaced by libertarian VCs or whatever nonsense that people are worried about.

Re: Emad Mostaque resigned as CEO of Stability AI

#219

Post by kalkr »

vishnugupta wrote: post #104
minimaxir wrote: post #3

He's leaving to work on decentralized AI? That's exactly what Stability AI was doing before it became clear the economics no longer work out in practice, and starting a new company wouldn't change that. (Emad is an advisory board member to a decentralized GPU company, though: https://home.otoy.com/stabilityai/ ) Obviously this is the polite way to send him off given the latest news about his leadership, but this rati…

"Decentralised" AI probably means something to do with crypto. Like mint this ocin to get access to that model? That's my guess.

decentralization != coinshit

Re: Emad Mostaque resigned as CEO of Stability AI

#220

Post by VladimirGolovin »

callalex wrote: post #89
roenxi wrote: post #43

Earlier quoted context omitted.

A game with no consistency in the art is probably enabled. We've crossed the threshold where something like Magic the Gathering could be recreated by a tiny team and a low budget. I don't think the limiting factor here is the software; it looks like we got AI-generated art pretty much as soon as consumer graphics cards could handle it (10 years ago it would have been quite hard). I'd be measuring progress in hardware…

Current AI is entirely incapable of generating the balanced and fun/engaging rule sets required for a MtG style game. Sure the art assets could be generated with skilled prompting and touchup but even that is nowhere close to the strong statement you made.

OP likely meant that a Midjourney-level AI can easily generate all the card art.

Obviously, current AIs cannot generate game rulesets because the game feel is an internal phenomenon that cannot be represented in the material domain and therefore AIs cannot train on it.

Post reply on HN