>"While the role of the artist isn’t going away soon, the role of stock image sites might disappear. "
Not, yet. While it's cheap relative to stock images, it's time consuming to generate exactly what you want. Prices for stock images will collapse for the common quick to use images but the price for the specialized high end images will hold their value or even increase in value. Those historical and such images will continue to be valuable.
It will be interesting to see if a specialized job will rise where people will get paid to generate just the right image. It might be called "A.I. image artist " This individual will generate an image with an A.I. but use graphic tools to finalize it for use.
Also there is nothing preventing Stock Image Sites themselves using Dall-E to generate additional images. Heck they can use their own existing images for training (which the other's can't due to copyright issues) to increase the portfolio, but the Stock Image Sites can access free public images.
So, counter-intuitively it may strengthen Stock Image Sites value
Considering how DALL-E is fed with so many stock images that it sometimes spontaneously generates specific stock image websites' watermarks on output,[1] this is the stupidest, blandest possible ouroboros.
The latent-diffusion[1] I've got running at home frequently generates stock image watermarks (e.g. "The London Skyline at night in the style of Carboni"[2], images 1, 2, and 6)
Just because "Stock Image Sites" haven't sued Dall-E, doesn't mean it is legal.
The minute Dall-E becomes a threat, these sites will enforce Dall-E to retrain from public / creative commons images. But Stock Image sites will and be one-up on Dall-E
It's almost you don't how to play chess or business strategy.
This is such a clever hack: “here, take this answer; it’s good the ‘AI’ generated it”, and if you can’t check like you did you might be able to spoof a self serving recommendation through simply because interprability of the inference is so difficult and reproducing might be costly.
That exception only applies when the license doesn't explicitly reserve the right. Open source licenses could be updated to prohibit use as training data, or at least to clarify the position.
open source isn't the same as public domain. the code is still copyrighted and has a specific license that has to be adhered to in order to be allowed to use it
Nobody knows whether or not training a NN is fair use. If it were ruled to be not fair use then it would basically shut down a large chunk
of ML research in the US, as it's not just Copilot that is training on copyrighted data. All the large language models require copyrighted data such as web crawls as there is just not enough public domain material around.
So even if you are right, and courts rule that training is not fair use, it seems likely that all the big tech companies would lobby Congress to restore the prior state of affairs, as not having an ML industry is somewhat of a national security issue if the rest of the world is going full steam ahead on making Skynet.
„Dad, Dad! Can you give me your gun? I want to shoot myself in my foot!“ - „Oh no, why would you ever want to do this?!“ - „the neighbours are doing it too!“ - „oh, alrighty then!“
I swear, world politics and economics isn’t much different from that, intellectually….
The problem with this analogy is that the companies are large enough to have leverage over the legal system to likely be able to avoid any consequences. Even if case law eventually rules in favor of the original copyright holders of the training data, it's the customers of the ML companies who would most likely be directly liable for infringement, not the ML companies themselves (though the customers could then try to sue them for damages).
Since there is no legal precedent and the law itself isn't clear about this use case, it's basically a huge gamble on a legal gray area at this point. For VCs the risk doesn't really matter as ML startups only need to exist long enough to provide an exit with high ROI and for enterprise companies it doesn't matter as ML products are just one of many ventures for them.
It's worth noting that unlike Germany, where book and newspaper publishers have won rather unusual copyright claims against companies like Google, in the US the big publishing industries to worry about are movies and music, and most ML projects right now seem to focus on generating images or text rather than music or video. If "AI generated music" caught on like DALL-E 2 did, I think we'd see a lot more contention over how copyright law applies to ML training data.
We could you know… ask people to donate content to these systems. They could train on Creative Commons and ask Twitter to build image license options in to their UI, then train on all the freely licensed images.
So we could actually follow copyright laws and still have an ML industry. But I’m not sure big tech wants to ask the public for consent. They would rather have free reign to do whatever they want.
By the way I don’t like copyright law or the concept of IP. But I find it a little annoying that I’m supposed to respect IP law and ML stuff can just ignore it. Also if big tech was forced to encourage people to share stuff with an open license, this would be a huge net good for society! Instead nothing changes but big tech gets to take advantage of peoples copyrighted works and artists can’t do anything to stop it. That kinda sucks.
As a tangent... a human being who reads up on websites (that are not explicitly in the public domain) is in essence also "training" themselves on that data. It would be really short-sighted (and impossible to start with) to forbid an AI to read it if a human is allowed..
But even human-generated work has all sort of domain-specific “fair use” rules to comply with, including plagiarism (academia), open source licensing (code), attribution for generative works (CC) and so on. People who make too closely generative works are scrutinized and face social and business pressures (art, law).
Today’s ML output throws everything into a mixer and then blanket-calls it ML-generated output, because the original training content has been separated from the social and legal frameworks that govern it.
Absolutely, but it's the output of the human (or AI) that plagiarizes that might be ethically wrong or illegal when used in certain ways, not the human (or AI) reading the input documents in the first place.
Eventually, like in the music industry, we will have dedicated legal teams taking anyone to court who is possibly using a combination of those 5 colors out of the 50 million color combinations we "own" simply because filling your work schedule with court cases is the right thing to do.
The funniest ones will be where ML independently reproduces a picture from unrelated materials with humans making a futile effort trying to figure out how it obtained this result.
What about my biological intelligence? Let's say I'm just training myself (learning) by browsing and reading a lot of open source projects, not even bothering to check what the licenses say. Am I possibly violating any license? And I will produce some output based on this, writing code at work (that output will not be copy-pasted, but certainly based on the "training" I received from reading through the open-source code).
Ah yes, I envision a language where every method is someones property with various subscription levels for each. Imagine how well maintained everything would be, how rich in features....
Agreed - having played around with DALL-E 2 a fair bit and having made a lot of usage of stock images over the years (for blog posts with specific subjects), I would say the former takes more work/time than the latter. With stock images I can just do a quick search on Shutterstock and find a lot of high quality options (usually), whereas with DALL-E 2 I need to figure out the exact prompt I want and iterate on it for a while. Stock images are not that expensive -- if you buy many it's as low as $2 per image, or on the high end (if you pay to just download a few per month) it's more like $10 per image. It does cost more, but time is money, so...
Time is only money if you someone will pay you for your time. If that’s not true, then it’s just a good excuse to spend your money when you could (in some cases should) be spending your time.
True - I guess I meant "time is money" in the broader sense really (ie I have other things I would prefer to spend my time doing, whether paid or not).
-- I agree with you - however it's not that time consuming to get what you want - it's pretty easy once you get used to DALL-E - so far there isn't anything I've not been able to get on a couple tries (granted after spending ~$30 learning the prompt system)- however once you're used to it - it's fairly easy - I agree that the market for very custom work will go through the roof - but "I need a burger" or "I need an American looking hot dog" eeeek!!! =) --
> It will be interesting to see if a specialized job will rise where people will get paid to generate just the right image.
I don't think so. People just need the result, so the AI will simply become a tool of the trade and you won't have any more AI image engineers than you have dedicated Photoshop artists right now.
I think the point of differentiation is that being adept at Photoshop is a relatively advanced and specialized skill.
Manipulating an AI prompt to get what you want is also a specialized skill, but may require an order of magnitude or two less training, or obviate the need for the job entirely.
An art director for a campaign wants a set of images created, and a set of stock photos used. They may have a junior person on their team create those images, each of which could take hours to create, or they can produce those images with an AI tool, which might take minutes. Or the director may simply use the tool themselves for a few minutes and then hand it off to someone else to clean up "in post".
Think people who think this is a job will be deluding themselves, eventually the magic of getting a good image or a bad image will just be a set of known good styles it generates at the start that you can then pick and choose from in the backend invisible to the actual thing you type. Same way how Dall-e 2 solves it's diversity issues.
The goal of OpenAI isn't to build a whole new industry of AI Artists, it's to make the AWS of creativity, which means it has to be so simple that you don't even need anyone who can write a good sentence. Just has to get good results from whatever they type.
Help me out. What's a "prompt engineer?" It seems like a very good title for the way we will be using AIs in the future. Many of these AIs will need just the right prompt (question?) to get the info we need. I like the title.
It's basically a way in which already delusional people are further deluding themselves that writing a part of a sentence while interacting with a black box is engineering.
Engineering is full of highly specialized disciplines. I met radio antenna engineers who only ever worked on one single piece of test equipment their entire working career and who used it to watch a signal bounce while they moved around pieces of copper tape on a PCB. That's literally all they did, and they didn't know what a spectrum analyzer or any other related tool was.
You'll find these specialized positions working with AI as well soon.
We can deride it from not being "traditional" engineering, but engineering as defined as "doing the best you can with tools you have", certainly allows for a prompt engineer to be defined..
>engineering as defined as "doing the best you can with tools you have"
This is not a definition of engineering anyone sane holds. It's so broad it can be used to define a floor cleaner as an engineer (or, if you want higher tech involved, a floor cleaner with a roomba).
>Prices for stock horses will collapse for the common quick to use horses but the price for the specialized high end horses will hold their value or even increase in value.
This was all true when cars became a thing. What’s the market cap of horse production companies before and after?
You are right. Humans will move on to building more high quality images. But for regular run of the mill stock images, AI is already there. I had done a small experiment to create stock images[1]
[1] https://medium.com/ozonetel-ai/generating-a-landing-page-wit...
OTOH, sometimes stock pictures made by humans aren't even that relevant if the entity using them is mostly just looking for filler: https://www.reddit.com/r/weirdwikihow/
This means the result in the best-use case will be developing a suitable in-house pattern language. I'm sure Adobe's all over this since it's related to language arts and will be a more natural fit for the critical eye of a design & branding team.
And probably still, downstream designers are going to be showing how they can convert DALL·E 2 imagery to polished finals. Especially after reviewing the blog post, it's really clear that if you want things to come together well for a refined corporate environment you'll need someone doing that. "I love the whale imagery, but I don't like the DALL·E 2 look, what can we do about the whale teeth" or whatever will definitely be a thing.
Not an artist, but (confirming against DeviantArt) with DALL-E etc it appears much lower cost to take bigger -- actually artistically more impressive -- risks. And dare I say more ego-less?
One thing I have wondered: what will these generators do to the corporate art market -- art bought in bulk for a hospital or office space. Will interior design specialists pick up prompt generation as an added skill?
Not, yet. While it's cheap relative to stock images, it's time consuming to generate exactly what you want. Prices for stock images will collapse for the common quick to use images but the price for the specialized high end images will hold their value or even increase in value. Those historical and such images will continue to be valuable.
It will be interesting to see if a specialized job will rise where people will get paid to generate just the right image. It might be called "A.I. image artist " This individual will generate an image with an A.I. but use graphic tools to finalize it for use.