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

Stable Diffusion Public Release

stability.ai

Post reply on HN
Print view View original

81–90 of 437 posts

Re: Stable Diffusion Public Release

#81

Post by dbingham »

wwwtyro wrote: post #51
mabbo wrote: post #33

The most interesting part, to me, of a release like this is the amount of "please don't abuse this technology" pleading. No licence will ever stop people from doing things that the licence says they can't. There will always be someone who digs into the internals and makes a version that does not respect your hopes and dreams. It's going to be bad. As I see it, within a couple years this tech will be so widespread and…

> Politics will be influenced. You can't trust anything you see anymore I've been wondering for a while now if this will lead to an unexpected boon: perhaps people will be forced to pay attention to a speaker's content instead of simply who is speaking.

The problem with this is that you will never know who is actually speaking. Deep fakes are already a thing, but as they get better and more accessible we will approach a world where anyone can make anyone say anything and make it hyper believable. In that world, it will be very difficult to tell what is real.

Re: Stable Diffusion Public Release

#82

Post by tastemykungfu »

suby wrote: post #73
gjsman-1000 wrote: post #55

Earlier quoted context omitted.

I had a thought for a utopian/dystopian world. One day these image generators will also get video support... and pornography support. When that happens, a few things may occur that I think are reasonable to predict: EDIT: Original post was way too wordy, TL;DR: When AI-generated pornography becomes available, it could be likely that demand for "real" pornography disappears because the AI will match and surpass the "r…

There will be a huge and unkillable market for non-AI generated pornography, even if people cannot tell the difference in an AB test. The demand will be too strong and I don't think there will be much outcry to ban it if it's all consenting adults.

Curious as to how you arrived at that conclusion.

Re: Stable Diffusion Public Release

#83

Post by soulofmischief »

wwwtyro wrote: post #56
daenz wrote: post #30

Earlier quoted context omitted.

>This is one of the most important moments in all of art history. I agree, but not for the reasons you imply. It will force real artists to differentiate themselves from AI, since the line is now sufficiently blurred. It's probably the death of an era of digital art as we know it.

Yeah, the fact that these models are necessarily based on existing works leaves me hopeful that humans will remain the leaders in this space for the time being.

Human works are needed to create the initial datasets, but an increasing amount of models use generative feedback loops to create more training data. This layer can easily introduce novel styles and concepts without further human input.

The time is coming where we will need to, as patrons, reevaluate our relationships with art. I fear art is returning to a patronage model, at least for now, as certainly an industry which already massively exploits digital artists will be more than happy to replace them with 25% worse performing AI for 100% less cost.

Re: Stable Diffusion Public Release

#84

Post by causi »

OmarIsmail wrote: post #72
mabbo wrote: post #33

The most interesting part, to me, of a release like this is the amount of "please don't abuse this technology" pleading. No licence will ever stop people from doing things that the licence says they can't. There will always be someone who digs into the internals and makes a version that does not respect your hopes and dreams. It's going to be bad. As I see it, within a couple years this tech will be so widespread and…

We do a decent job of banning child pornography. And bringing the two ideas together, is child pornography that is provably created by an AI still illegal?

As far as I'm aware countries fall broadly into two camps. Camp 1, USA for example, is concerned purely with the abuse of children, i.e., anything that depicts or is constructed of pieces of real children is illegal but other things such as drawings, stories, adults role playing, etc is not. Camp 2 outlaws any representation of it whether or not a child was involved.

Nowhere will a training set featuring pictures of naked children be legal.

Re: Stable Diffusion Public Release

#85

Post by mkaic »

egypturnash wrote: post #41
mkaic wrote: post #2

This is one of the most important moments in all of art history. Millions of people just got unconditional access to the state-of-the-art in AI text-to-image for absolutely free less the cost of hardware. I have an Nvidia GPU myself and am thrilled beyond belief with the possibilities that this opens up. Am planning on doing some deep dives into latent-space exploration algorithms and hypernetworks in the coming days…

It’s the opposite of exciting if art is your job, lemme tell you.

(OP here) I agree. I am an artist (not by trade but by lifelong obsession) in several different mediums, but also an AI engineer--so I feel a weird mixture of emotions tbh. I'm thrilled and excited but also terrified lol.

Re: Stable Diffusion Public Release

#86

Post by topynate »

bluejellybean wrote: post #39

While neat, and no doubt impressive, it still utterly fails on prompts that should be completely reasonable to any sane human being/artist. Take something like "A cat dancing atop a cow, with utters that are made out of ar-15s that shoot lazer-beam confetti". A vivid description should be aroused in your head, and no doubt, I could imagine an artist have a lot of fun creating such a description... Alas, what the mode…

The referent of "utters" (sic) is ambiguous, so I can imagine a model having more difficulty with it than usual. Regardless, the current SOTA does need more specific and sometimes repetitive prompting than a human artist would, but it's surprising how much better results you can get from SOTA models with a bit of experience at prompt engineering.

Re: Stable Diffusion Public Release

#87

Post by mkaic »

tailspin2019 wrote: post #15
mkaic wrote: post #2

This is one of the most important moments in all of art history. Millions of people just got unconditional access to the state-of-the-art in AI text-to-image for absolutely free less the cost of hardware. I have an Nvidia GPU myself and am thrilled beyond belief with the possibilities that this opens up. Am planning on doing some deep dives into latent-space exploration algorithms and hypernetworks in the coming days…

> state-of-the-art in AI image-to-text I think you meant text-to-image!

oh heck, you're right! I edited the original comment, thanks.

Re: Stable Diffusion Public Release

#88

Post by alcover »

Moral question:

Such tool will be used to generate lewd imagery involving virtual minors.

No way to prevent it upstream by outlawing feeding it real content (whose possession already is illegal). Suffice to add 'childrenize' layer onto adult NN or something.

How will the legal system react ? Bundle it into illegal imagery, period ? Maybe it's already the case - I think drawing made public is, not sure. If not, on what ground could it be ? No real minor would be involved in that production.

Re: Stable Diffusion Public Release

#90

Post by narrator »

mabbo wrote: post #33

The most interesting part, to me, of a release like this is the amount of "please don't abuse this technology" pleading. No licence will ever stop people from doing things that the licence says they can't. There will always be someone who digs into the internals and makes a version that does not respect your hopes and dreams. It's going to be bad. As I see it, within a couple years this tech will be so widespread and…

As someone who witnessed AI Dungeon's GPT-3 model with an unlimited uninhibited imagination for erotica I would download everything now before they cripple the models. I would not be surprised if they very shortly completely stop downloads due to "abuse" and pursue a SAAS model.

I think it's funny how Yandex, a Russian company, releases these big language models without all the AI safety handwringing in their press releases. The Russians have a tradition of releasing technology without giving a lot of worry about what happens to it, for better or for worse. For example, they made between 75 and 100 million ak-47s, a not in any way limited machine gun, and it spread to every corner of the world. They even gave out all the plans and technical assistance so any one of the organizations they worked with could produce their own. 20 different countries make ak-47s currently. Of course you had to register every xerox machine in the Soviet Union, so maybe they just had different priorities?

The west is absolutely fascinated these days with the control of advanced technology. Drones, Blockchain, and AI models seem to be the latest things that the west is determined to exercise control over. For example:

"Many of the technological advances we currently see are not properly accounted for in the current regulatory framework and might even disrupt the social contract that governments have established with their citizens. Agile governance means that regulators must find ways to adapt continuously to a new, fast-changing environment by reinventing themselves to understand better what they are regulating. To do so, governments and regulatory agencies need to closely collaborate with business and civil society to shape the necessary global, regional and industrial transformations." -Klaus Schwab, "The Fourth Industrial Revolution", Page 70.

Post reply on HN