Is AI Art Having its Napster Moment?
Just this week, Getty Images sued the makers of Stable Diffusion, an artificial intelligence system (“AI”) that creates art based on existing art work, like Getty’s. Long before this copyright law suit, the line between artistic “influence” and infringement had been a blurry one, but some think the very the paper it’s written on is disappearing.
The U.S. began losing its grip on copyright with an earlier groundbreaking technology heralded by Napster, the “file sharing” (and copyright infringing) web application that seemingly everyone had in the early 2000’s. Millions of ordinarily law-abiding citizens in developed countries began stealing music electronically without a moment’s thought. Easy Internet file sharing was a perfect storm of technological progress and easy moral degradation. Never before was it so simple to steal, aided and abetted by our laptops and the Internet, thanks to this incredible new technology.
And the music industry has never been the same since. Today, artists don’t sell albums, so much as engage in an unartistic daily grind of exploiting their persona, with elaborate “360 contracts,” product promotion, and relentless playing out in dive bars to make a living. As a result, in all of the arts, there has been a recent decline in creativity, by many measures.
Might AI-art be the next chapter in technology-fueled erosion of artist’s rights, where a promising new technology winds up disrupting, not enhancing, creativity? Or will it simply replace our artists altogether?
AI is just starting to surprise us with its powerful capabilities, and nowhere more so than in the arts. We thought that art would be the last realm to fall to the robots – too complex, too inherently human an endeavor, for AI.
We were wrong there.
Art is the first major battlefield with AI, not our last stand. And, as with Napster, it’s an uphill battle. Art is highly subjective, often abstract, feeding on complex patterns and chaotic new forms. It is a far easier AI target than, say, the unforgiving perfection and correctness required for robotic surgery.
Just like with Internet file-sharing, the easiest place for this new technology to start is with what is already proven – the art we already love.
So that’s what Stable Diffusion does. It takes several existing first-in-class source images, like Getty’s, and applies an ever-improving AI algorithm to generate “new” art. New enough to arguably be “influenced by,” but not infringe, the copyrights of those sources. Its’ machine learning algorithms can’t be so new that they’re unrelatable and unappealing. So, like Internet file sharing, AI art is exceedingly likely, by design, to step on the blurry line of copyright infringement.
But as with Internet file sharing, will people simply not care, because they are getting so much free, easy art? We already have hundreds of AI Art projects, similar to Stable Diffusion, and people are raving about them. Could we even stop this if we wanted to, with infinite infringing artworks churned out every minute?
Will copyright’s “blurry line,” and the digital paper its written on, simply fade away altogether, into the infinite void of AI art?
We think that overstates the case, but, as with Napster, artistic industries will have to transform themselves to swim in this new, endless sea of unlimited “AI artists.”
