I’ve gotten kind of obsessive about something lately.
AI companies fed vast amounts of copyrighted work into their technology to create the AI we have today. All this creative work has provided economic value to the booming AI companies, but the creators of the original work weren’t compensated. Shouldn’t that violate copyright law? Well, I think so. But the Courts are still deciding, and the loudest voices are suggesting it was totally fine for them to do this, as if there was no other way to develop such a revolutionary technology. I don’t think that’s quite right.
When I was invited to adapt (into comics) an AI vision paper called Shaping AI this year, I noticed that the paper’s discussion of AI’s potential impact on media and entertainment had overlooked this issue. If AI is so smart, can’t it be applied to solve the problem of artists not being compensated (or even asked) when their copyrighted work was used to train AI models?
I was pleasantly surprised when the authors took my feedback and highlighted this idea as one of the milestones to focus research on and try to make possible. The illustrations in this post are my adaptation of that section of the paper.
I believe that if we decide that it is a problem worth solving, it would be possible to develop AI in a way that allows artists and writers to be compensated for this new, additional form of value that their work has, as “AI food.”
What I don’t know yet is how to make that happen.1
Thanks for being here!