Shutterstock Launches Generative AI Image Tool (gizmodo.com) 34
Shutterstock, one of the internet's biggest sources of stock photos and illustrations, is now offering its customers the option to generate their own AI images. Gizmodo reports: In October, the company announced a partnership with OpenAI, the creator of the wildly popular and controversial DALL-E AI tool. Now, the results of that deal are in beta testing and available to all paying Shutterstock users. The new platform is available in "every language the site offers," and comes included with customers' existing licensing packages, according to a press statement from the company. And, according to Gizmodo's own test, every text prompt you feed Shutterstock's machine results in four images, ostensibly tailored to your request. At the bottom of the page, the site also suggests "More AI-generated images from the Shutterstock library," which offer unrelated glimpses into the void.
In an attempt to pre-empt concerns about copyright law and artistic ethics, Shutterstock has said it uses "datasets licensed from Shutterstock" to train its DALL-E and LG EXAONE-powered AI. The company also claims it will pay artists whose work is used in its AI-generation. Shutterstock plans to do so through a "Contributor Fund." That fund "will directly compensate Shutterstock contributors if their IP was used in the development of AI-generative models, like the OpenAI model, through licensing of data from Shutterstock's library," the company explains in an FAQ section on its website. "Shutterstock will continue to compensate contributors for the future licensing of AI-generated content through the Shutterstock AI content generation tool," it further says.
Further, Shutterstock includes a clever caveat in their use guidelines for AI images. "You must not use the generated image to infringe, misappropriate, or violate the intellectual property or other rights of any third party, to generate spam, false, misleading, deceptive, harmful, or violent imagery," the company notes. And, though I am not a legal expert, it would seem this clause puts the onus on the customer to avoid ending up in trouble. If a generated image includes a recognizable bit of trademarked material, or spits out celebrity's likeness -- it's on the user of Shutterstock's tool to notice and avoid republishing the problem content.
In an attempt to pre-empt concerns about copyright law and artistic ethics, Shutterstock has said it uses "datasets licensed from Shutterstock" to train its DALL-E and LG EXAONE-powered AI. The company also claims it will pay artists whose work is used in its AI-generation. Shutterstock plans to do so through a "Contributor Fund." That fund "will directly compensate Shutterstock contributors if their IP was used in the development of AI-generative models, like the OpenAI model, through licensing of data from Shutterstock's library," the company explains in an FAQ section on its website. "Shutterstock will continue to compensate contributors for the future licensing of AI-generated content through the Shutterstock AI content generation tool," it further says.
Further, Shutterstock includes a clever caveat in their use guidelines for AI images. "You must not use the generated image to infringe, misappropriate, or violate the intellectual property or other rights of any third party, to generate spam, false, misleading, deceptive, harmful, or violent imagery," the company notes. And, though I am not a legal expert, it would seem this clause puts the onus on the customer to avoid ending up in trouble. If a generated image includes a recognizable bit of trademarked material, or spits out celebrity's likeness -- it's on the user of Shutterstock's tool to notice and avoid republishing the problem content.
Ready, fire, aim. (Score:1)
So many questionsâ¦
Enforcement of the âoedonâ(TM)t be badâ agreement? Anyone willing to go north of 0.001% effectiveness on this?
Would make a dandy plan to generate lotsa work for lawyers.
Whatâ(TM)s to anyone from scraping lotsa sample images and using the open source process to do the same thing? Sure, they wonâ(TM)t be high def, but prob good enough for effective mobile web resolution.
Re: (Score:2)
There's much bigger questions, like who wants to buy a can of "PESPISII" from a 7-fingered mutant, created by a person with no eye for art?
AI image generation tools rarely spit out professional quality images on their own, even if you have a lot of experience with prompting. You still need a person who knows what they're doing doing cleanup, bare minimum. Also, it's not like hiring artists is exactly expensive, so whose time is better spent generating and cleaning up images, an artist earning $15-$20 an h
Re: (Score:3)
The entire point is that instead of having 1000 of artists generating 1000s of images from scratch, you will now have 10 artists fixing the minor flaws in AI's work.
It's not removing artists. It's making them far more efficient. Same thing as happened with accountants and Excel.
Re: (Score:2)
Yes and no, because then you hit Jevons Paradox [wikipedia.org].
Re: (Score:2)
Except that AI has near infinite scaling, so scaling will be done via AI, not artists.
Re: (Score:2)
What you're talking about requires AGI. We're not even anywhere near your proposed 100:1 ratio (honestly, not even 10:1), let alone "infinite".
And if we have AGI, you have bigger things to worry about. Namely because then you're either going to be (A) lounging around on a beach while a benevolent AI attends to your every need and desire, or (B) hiding in the woods eating bugs while a malevolent AI hunts down and exterminates the last remnants of humanity.
Re: (Score:2)
This requires self aware AI as much as accounting work done on Excel requires self aware AI.
This is why I put a qualifier of "near" before "infinite". Again, this isn't some kind of hypothesis. We're already starting to see this at the gaming companies, who're investing hard into their own art AI. Because that would massively reduce the amount of artists they need to hire. And artists are really hard to keep filled with work in gaming development cycle, because once you have art done, it's actually done and
Re: (Score:1)
Re: (Score:3)
Yes, minor flaws. Because they're easily corrected by a human. And then the corrections can keep teaching the AI.
Large gaming companies are already working on this in house. Take their art style, have AI do most of the work, and have artists correct the warts. Work that used to require tens to hundreds will be done by a handful in the future once training is complete.
Called it (Score:2)
I mean this probably isn't a surprise to anyone, and it was no shocking prediction, but I've been saying we're rapidly heading towards a future where you don't use Google Images or stock photo services. Instead you'll type what input images you need for your project and an AI generates exactly what you want. Then it cuts it out for you into transparent layers, and even lets you do a rough collage and iterate over that as a rough input image, to generate one with all the edges nearly composed and blended and
The horror, the horror! (Score:1)
Cool... (Score:3)
Re: (Score:2)
The cost of these models isn't too bad. It's pretty good at letting you get cheap illustrations of whatever you want.
For a marketing plan, you could use generated pictures as stand in before the real photos are there.
For video game purposes, you can generate stand-in sprites of characters before really knowing enough about what you want to pay an artist.
I've used some of them to generate ideas of logos for a team we were building at the time.
Might not generate the final version of what you want, but you can
Re: Cool... (Score:1)
How does it compare to something like Photoshop? I'm guessing that the generator's finer points need explaining as they miss me....
Re: (Score:2)
Sadly, most of what it gets used for is porn :P
Compensate Photographers for Photos Used - BS (Score:2)
This is like paying people to train their replacements in India. I doubt there was any clause in the contracts to use AI to copy aspect of their photos. I hope they get sued to oblivion.
Re: (Score:2)
I asked an AI to draw a picture of Chuck Norris (Score:2)
Next moment he is on my doorstep telling me to quit the nonsense as there are no datasets large enough to capture the brilliance of Chuck.
I'll stick to unicorns as there are many picture of those for the AI to learn from.
What's the point (Score:2)
There's a Musician/YouTuber Benn Jordan. Who points out why be afraid of ai generated images and music, when there are already sites like pexels and unsplash that have millions of free images and music...
https://youtu.be/Jb7KD1WaX7o [youtu.be]
So what are they trying to achieve? Cheaper images than free? Images without copyright?
If I were shutterstock... I'd be worried that people would start to associate my brand with ai/fake images... Instead of real life things... People probably won't pay for fake? Or like being sh
Re: (Score:2)
Re: (Score:2)
Why exactly did you link an anti-pesticide ad?
Re: What's the point (Score:1)
Sorry. I must have copied the link of the advert... Not the video. Here's the actual link
https://youtu.be/YcvXNdkKQxE [youtu.be]
Re: What's the point (Score:2)
Parody and Protest Placards (Score:2)
Re: What's the point (Score:1)
Sorry the link above is wrong... But his name isn't, search for Benn Jordan.. Its one of his more recent vids..
https://youtu.be/YcvXNdkKQxE [youtu.be]
Even with "Licensed Dataset" (Score:1)
Re: (Score:2)
This is a misrepresentation of the current (limited) cases. Copyright requires human creative endeavour. The initial case that kicked this all off was when a person tried to copyright, in a machine's name, a creation the machine created with no human prompting. Which failed for a lack of human creative endeavour.
The situation is far from settled, but I strongly suspect that where this is going to end up at is basically, "What did you do before and after clicking "Generate", and how much creativity was ther