Nice article Fawzi. I came to a similar conclusion a while back in a piece I wrote, but you've fleshed it out in a lot more detail here. Great to see :)
"I think the only truly ethical way forward for AI-imagery, is to create an open training data set that people have to explicitly opt-in to. The financial side would need to be worked out, but perhaps there are small payments for initially submitting images to be used in training data, and maybe some kind of royalties system based on prompts that explicitly use artists names (and much, much higher royalties for anything used in commercial work)? This could be scaled with some kind of CC-BY type license that is explicitly for use in AI training data, exposed through image metadata somehow."
I still wonder how many artists would opt-in and how much money they would make if they did. But I'm glad that Shutterstock is exploring the more ethical path as opposed to keeping profits to themselves.
That is interesting. I still worry a lot about opt-out vs opt-in systems, and especially platforms with a ton of images already (instagram, Shutterstock, etc) who likely have gotten permission to use their image libraries for whatever, but not explicitly for AI training. We'll definitely see how it plays out, but I assume the result will be somewhat similar to Spotify and similar models where "whatever makes VC's and corporations the most money" will likely win out. Interesting times!
That's my biggest worry, especially with Instagram. Zuck just announced that they were creating their own generative AI tools and I can't help but think of them using every image on IG as training data because some vague clause in the T&C's covers it.
You make an interesting comparison to Spotify. I see a lot of parallels. AI art generators are here to stay, and artists might have to embrace the movement (even if they're not compensated fairly) instead of fighting it. I think it would be great if these tools implemented a discovery tool that connected people to artists for more serious work, similar to what I showed above.
Jan 28, 2023·edited Jan 28, 2023Liked by Fawzi Ammache
I like the ArtFair concept, but the Genie has been let out. The distributive technology for $ would be easy now, but the will to spread the wealth just is not there. The problem I can quickly prompt: /imagine (AI pun) is access to AI becoming limited to the point there becomes a new digital divide. New layers of us and them.
I think AI will be accessible to all (or at least affordable to most people), but we'll start seeing the usual trend of creating "tiers" for more premium features. An example of this is the rumoured "ChatGPT Pro" that could cost up to $42/month. I don't think this is a bad thing as long as there are ranges of options that allow everyone to participate.
Jan 27, 2023·edited Jan 27, 2023Liked by Fawzi Ammache
Art Fair is an interesting concept. But I doubt any AI developer would be willing or able to invest millions or billions of dollars up front to train their software. And you would have to sign up a crazy number of artists for it to work. I think it is more likely that AI art will simply be art used as the final artwork. I see an increasing number of writers using it for story images already. It will improve with time. And so will writing AI. More creatives will need day jobs.
Yep, ArtFair is definitely idealistic. I think AI developers would have leverage since they're paying out royalties based on how much an artist contributes, which could make them sell it for a reasonable price. And the platform would give them access to a network of customers who might be willing to pay a bit extra for well-polished art.
I think one-off creations using AI are still good enough in some situations. Like you said, many writers have been using them for their blog posts. Although in this particular situation, I don't think writers would have hired artists anyway. The AI-generated art replaced the free stock images they used to get from sites like Unsplash.
That is true. But once AI illustrations are up to snuff and easy to produce, professional media and print editors will be using them also. Like they already do photography from UnSplash. There will still be small markets for illustrators, like children’s books or comic books that need a lot of related content, but much single image work will go to AI.
Love this concept. ✨
Thanks Elle!
Nice article Fawzi. I came to a similar conclusion a while back in a piece I wrote, but you've fleshed it out in a lot more detail here. Great to see :)
"I think the only truly ethical way forward for AI-imagery, is to create an open training data set that people have to explicitly opt-in to. The financial side would need to be worked out, but perhaps there are small payments for initially submitting images to be used in training data, and maybe some kind of royalties system based on prompts that explicitly use artists names (and much, much higher royalties for anything used in commercial work)? This could be scaled with some kind of CC-BY type license that is explicitly for use in AI training data, exposed through image metadata somehow."
https://clipcontent.substack.com/p/the-ai-art-debate-excitement-fear
Thanks for sharing, Steve! I'm glad I don't sound crazy and that other people have similar ideas 🤣
After writing this piece, I learned that Shutterstock will be following a similar approach to paying artists: https://techcrunch.com/2023/01/25/after-inking-its-openai-deal-shutterstock-rolls-out-a-generative-ai-toolkit-to-create-images-based-on-text-prompts/
I still wonder how many artists would opt-in and how much money they would make if they did. But I'm glad that Shutterstock is exploring the more ethical path as opposed to keeping profits to themselves.
That is interesting. I still worry a lot about opt-out vs opt-in systems, and especially platforms with a ton of images already (instagram, Shutterstock, etc) who likely have gotten permission to use their image libraries for whatever, but not explicitly for AI training. We'll definitely see how it plays out, but I assume the result will be somewhat similar to Spotify and similar models where "whatever makes VC's and corporations the most money" will likely win out. Interesting times!
That's my biggest worry, especially with Instagram. Zuck just announced that they were creating their own generative AI tools and I can't help but think of them using every image on IG as training data because some vague clause in the T&C's covers it.
You make an interesting comparison to Spotify. I see a lot of parallels. AI art generators are here to stay, and artists might have to embrace the movement (even if they're not compensated fairly) instead of fighting it. I think it would be great if these tools implemented a discovery tool that connected people to artists for more serious work, similar to what I showed above.
I like the ArtFair concept, but the Genie has been let out. The distributive technology for $ would be easy now, but the will to spread the wealth just is not there. The problem I can quickly prompt: /imagine (AI pun) is access to AI becoming limited to the point there becomes a new digital divide. New layers of us and them.
I think AI will be accessible to all (or at least affordable to most people), but we'll start seeing the usual trend of creating "tiers" for more premium features. An example of this is the rumoured "ChatGPT Pro" that could cost up to $42/month. I don't think this is a bad thing as long as there are ranges of options that allow everyone to participate.
Art Fair is an interesting concept. But I doubt any AI developer would be willing or able to invest millions or billions of dollars up front to train their software. And you would have to sign up a crazy number of artists for it to work. I think it is more likely that AI art will simply be art used as the final artwork. I see an increasing number of writers using it for story images already. It will improve with time. And so will writing AI. More creatives will need day jobs.
Yep, ArtFair is definitely idealistic. I think AI developers would have leverage since they're paying out royalties based on how much an artist contributes, which could make them sell it for a reasonable price. And the platform would give them access to a network of customers who might be willing to pay a bit extra for well-polished art.
I think one-off creations using AI are still good enough in some situations. Like you said, many writers have been using them for their blog posts. Although in this particular situation, I don't think writers would have hired artists anyway. The AI-generated art replaced the free stock images they used to get from sites like Unsplash.
That is true. But once AI illustrations are up to snuff and easy to produce, professional media and print editors will be using them also. Like they already do photography from UnSplash. There will still be small markets for illustrators, like children’s books or comic books that need a lot of related content, but much single image work will go to AI.