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

391–400 of 429 posts

Re: Emad Mostaque resigned as CEO of Stability AI

#391

Post by __loam »

JCM9 wrote: post #330
ASinclair wrote: post #323

Earlier quoted context omitted.

> Business minded folks are already starting to deeply question where the value is relative to the amount of money spent on training. Not just training but inference too, right? They have to make money off each query.

Yes. Each query needs to generate enough revenue to pay for the cost of running the query, a proportional cost of the cost to train the original model, overhead, SG&A, etc. just to break even. Few have shown a plan to do that or explained in a defensible way how they’re going to get there. A challenge at the moment is a lot of the AI movement is led by folks that are brilliant technologists but have little to no expe…

Many here know I personally despise AI, but to take off my tech critic hat for a moment, if subscription models don't work out, we still ultimately do have a bunch of pretty powerful base models to build on. If you can push inference to the edge and run on local machines, you can make your users bear the cost.

Re: Emad Mostaque resigned as CEO of Stability AI

#392

Post by teaearlgraycold »

BhavdeepSethi wrote: post #191
alfiedotwtf wrote: post #168

Earlier quoted context omitted.

Sometimes you need to say fuck the money, I’ve already got enough, and I just want to do what I enjoy. It may not be an ideal model for HN but damn not everything in life is about grinding, P/E ratios, and vesting schedules

Yeah, that's easier to say when you have enough. A lot of employees might not be in that privilege position. The reality for some of the folks might be addressing education loans, families to take care of, tuition for kids, medical bills, etc.

I’m grateful to be debt free

Re: Emad Mostaque resigned as CEO of Stability AI

#393

Post by cchance »

JCM9 wrote: post #330
ASinclair wrote: post #323

Earlier quoted context omitted.

> Business minded folks are already starting to deeply question where the value is relative to the amount of money spent on training. Not just training but inference too, right? They have to make money off each query.

Yes. Each query needs to generate enough revenue to pay for the cost of running the query, a proportional cost of the cost to train the original model, overhead, SG&A, etc. just to break even. Few have shown a plan to do that or explained in a defensible way how they’re going to get there. A challenge at the moment is a lot of the AI movement is led by folks that are brilliant technologists but have little to no expe…

Inference will continue to shift toward profitability, Google and the rest are going to choke the cost to gain traction but as TPU costs come down and GPU costs scale up and models become more efficient the cost to infer will drastically scale down into profitability.

Re: Emad Mostaque resigned as CEO of Stability AI

#394

Post by cchance »

monkeydust wrote: post #224
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.

Reminded me of Petals offering distributed ML inference anyone tried this? https://github.com/bigscience-workshop/petals

yep was shockingly fast actually

Re: Emad Mostaque resigned as CEO of Stability AI

#395

Post by llm_trw »

raincole wrote: post #158
llm_trw wrote: post #138

Earlier quoted context omitted.

The `big' AI models are trillion parameter models. The medium sized models like GPT3 and Grok are 185b and 314b respectively. There is no way for _anyone_ to run these on a sub $50k machine in 2024, and even if you can the token generation speed on CPU is under 0.1 tokens per second.

It's just semantic gymnastics. I'm sure most people will consider LLaMa 70B a big model. Of course if you define big = trillion then sure big = trillion[1]. [1]: https://en.wikipedia.org/wiki/No_true_Scotsman

Yes, you are engaging in the no true scotsman fallacy, please stop.

Re: Emad Mostaque resigned as CEO of Stability AI

#396

Post by rationalfaith »

alephnerd wrote: post #319
JCM9 wrote: post #307

2024 is going to shift into a tough year for AI. Business minded folks are already starting to deeply question where the value is relative to the amount of money spent on training. Many/most of the GenAI companies have interesting ideas but no real business plan. Many of the larger AI companies look very shaky in terms of their governance and long term stability. Stability is showing itself to be very unstable, OpenA…

> Business minded folks are already starting to deeply question where the value is relative to the amount of money spent on training Kinda. In my experience, the bigger issue is the skillset has largely been diffused. Overtraining on internal corpora has been more than enough to enable automation benefits, and the ecosystem around ML is very robust now - 10 years ago SDKs like Scikit-learn or PyTorch were much less r…

[dead]

Re: Emad Mostaque resigned as CEO of Stability AI

#397

Post by numpad0 »

ben_w wrote: post #358
numpad0 wrote: post #305

Earlier quoted context omitted.

> Der Kuss (by Klimt) is widely regarded as being good art, The point is, generative AI images are not widely regarded as good art. They're often seen as passable for some filler use cases and hard to tell apart from human generations, but not "good". It's not not-there-yet because AI sometimes generates sixth fingers, it's something another level from Gustav Klimt, Damien Hirst, Kusama Yayoi, or the likes[0]. It cou…

> The point is, generative AI images are not widely regarded as good art. They're often seen as passable for some filler use cases and hard to tell apart from human generations, but not "good". > It's not not-there-yet because AI sometimes generates sixth fingers, it's something another level from Gustav Klimt My point is: yes AI is different — it's better . (Or, less provocatively: better by my specific standards ).…

> and yet to my eye it messes with anatomy

wait what? Isn't that missing the point of expressionism? Klimt's Judith I is basically a photo, surely he can draw sh*t if he wanted to?

But myriad predecessors such as Vermeer, Rembrandt, Van Gogh, da Vinci, et al., have done enough in realism, and also photography was becoming more viable and more prevalent, that artists basically started diversifying? Isn't that what lead to various forms of early 20th century arts like surrealism(super-real -ism), cubism, etc?

I don't mean offense but that's just, surely that level of understanding can't be basis of policy decisions when it comes to moral rights and licensing discussions and "artists should just use AI" and such???

Re: Emad Mostaque resigned as CEO of Stability AI

#398

Post by nine_k »

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.

It does not take a large team to generate interesting and balanced rules. I'd hazard to say that with a larger team the task becomes harder.

Creating the art, on the other hand, goes from a week per card to half a day per card, or something similar.

Re: Emad Mostaque resigned as CEO of Stability AI

#399

Post by hsjsbeebue »

immibis wrote: post #381
immibis wrote: post #190

Earlier quoted context omitted.

That's right. We invented programming AI a very long time ago, and called it an "assembler". All you had to do was tell the assembler what kind of program you wanted, and it would do the programming work for you! Then we invented another AI to tell the assembler what kind of program you wanted, and called it a "compiler". All you had to do was tell the compiler what kind of program you wanted it to tell the assembler…

P.S. Visual Basic with its GUI designer was a quite effective way to rapidly build apps of questionable quality but great business value. Somebody should bring that paradigm back.

Excellent as long as you jump ship when the complexity is over 5kloc

Re: Emad Mostaque resigned as CEO of Stability AI

#400

Post by johnnyanmac »

dagelf wrote: post #379
johnnyanmac wrote: post #202

Earlier quoted context omitted.

>It won't be decades, but a few more years would be helpful I see it this way to be honest: - companies will aggresively try to use AI in the next 2-3 years, downsizing themselves in the meantime - the 3-5 year launch mark will show that downsizing was an awful idea and took too many hits to really be worth it. I don't know if those hits will be in profits (depends on the company) but it will clearly hit that uncanny…

Models are already general, and you can set them to recursively self improve narrowly and still need humans in the loop for general improvement... but those humans are only going to get less... so change your years into months instead and multiply by rand(2)... and change the hire-backs into more startups doing more things...?

>so change your years into months instead and multiply by rand(2)... and change the hire-backs into more startups doing more things...?

I disagree completely. But I should note I was referring to medium to large scale companies. Nothing in those companies happens in "months" these days.

Maybe some startups rise from being first to market much faster, but given the huge legal issues I'm not seeing it. Microsoft et al. can afford A lengthy legal battle and make backup plans. A startup can't.

Post reply on HN