gizmodo.com: In the coming months, the stock image giant says it will expand its relationship with OpenAI to provide its users “direct access” to the wildly popular DALLE-2 text to image AI model. In effect, that means the two companies will work together to sell AI generated art trained, in part, from human produced Shutterstock stock images. Go ahead and insert your Ouroboros meme here.
6 comments:
I think that this could be a really cool concept for the use of this new AI generation. Although, I think that using other artists’ art in order to do it is exactly what Shutterstock should NOT be doing. The people they are compensating for their art don’t really have a choice whether or not their work actually gets used or not. There is no way to know whether the AI sucks in your image and uses it as inspiration or not. I think that maybe if Shutterstock HIRED people specifically for this project, that would be a different story. Instead, you are taking people’s work that they put on your website and using it to make your own. Even so, I feel like this AI art should be in a different setting; have its own tab on the website. I don’t think it should be mixed in considering people may want to use human-taken photos and support individual artists. I don’t think we should be making it even HARDER to support small artists.
Using artificial intelligence to create stock images is actually kind of genius in my opinion. Very often, I find myself sifting through so many stock images because I can’t find exactly what I want. With AI, you can create any kind of image you want, eliminating the need for the searching and the sifting. It is really interesting seeing how fast AI imagery has really taken over. I believe it was only a few months ago that the Dall-E image generator was released for public use, and already it’s infiltrated internet culture, art contests, and now stock images. I do agree that shutterstock jumped to commercialize this new technology very fast. That’s capitalism, though. I never really thought about the beef between the stock photo companies, Shutterstock and Getty Images really have it out for each other. I just know Getty is mad that Shutterstock got to the AI thing first.
2. This article brings up so many questions. I think the most important question to answer here is how are these AI images being generated? Are they using copyright free images as training material? In which case, would those humans actually be entitled to any compensation? If Dall-e is being fed images that are protected intellectual property, are those artists being paid for the use of their work in this algorithm? I think these areas of concern should be the primary target before we start considering the legality of monetizing what Dall-e creates. I think Shutterstock is indeed jumping in the deep end and should hold off. That being said, who is going to bring these concerns to court? The internet has always been purposefully a Wild West of minimal regulation. What will it take to force regulations to be created? Is that something that would be more beneficial in the end?
Honestly it is both terrifying and frightening that we find ourselves living in a time where this is happening. There are so many legal questions that we probably haven't even bothered to ask ourselves yet. The fact that we have created what is essentially just false imagination is wild to me. I mean, screw being able to type a desired image into Google or some other image search site, now it doesn't even matter if the image exists yet. We've created something that will try it's damnedest to create it for you, and it's actually getting pretty damn good at it. The implications of this tech is astounding. How long until all of those chatbot or AI story generators where you riff back and forth with the computer actually begin to generate compelling stories?
I'm excited to see where the future takes us, but I think we are probably going to look back and be like "damn, those things were so rudimentary in comparison to what we have going on now". This tech is the shoulders of giants that are future ai will be based on. it's incredible.
As a few other commenters have said, this article brings up a lot of questions, legal, moral and otherwise. While it’s pretty cool that AI can generate images to this degree of accuracy and quality, I don’t know if I’m really that big a fan of this plan by Shutterstock. With this deal a number of competitors to Dall-E have to deal with effectively being locked out from being able to compete with/offer their product to the general public on the same scale, which seems questionable to me. A comment on the original article that I thought was interesting said, in short, that this, selling AI generated photos partially trained from a free, open source web dataset, is peak late stage capitalism, and I don’t think that I necessarily disagree. This technology is an interesting one, and I’m curious to see how it progresses into the future - I’m sure there will be many questions as it continues to develop.
This is definitely a challenging topic to debate for sure. If a computer is generating the images then whose work is it really? I like the concept that Shutterstock is talking about with their contributor fund and the royalties each time an image from a creator is referenced. It sounds difficult to me to keep track of all of that so I hope they aren’t just trying to appease their creators by saying this. This is classic capitalist greed that we constantly see. Taking something that was open source and free to use and monetizing it for profit. I’m sure Shutterstock will make this some sort of subscription based service and make a ton of money off of this while the creators will end up making a fraction of that. All in all though, I think this technology is extremely impressive and useful in so many circumstances. I am already seeing it in a lot of my classmate’s visual research for shows and other situations like that. I hope that Shutterstock doesn’t ruin it.
Post a Comment