In recent years, artificial intelligence (AI) has been making inroads in virtually every industry. Chatbots exist for everything from cognitive behavioral therapy to translations to task management. Cars can park themselves. Some even drive themselves.
And while a decade or two ago the idea of AI creating art might have seemed purely science fiction, we now have AI-generated art not only being created but also being sold for prices that would make a lot of artists very jealous. Recently, Google’s AI-created art raised $84,000 in a two-day exhibit.
So what does all this AI art mean for human artists? Is AI eventually going to push humans out of the art world all together? Can AI even create original art that humans would find attractive? And who owns the art AI creates—the AI or the person who created?
Can AI Create Attractive Art?
Is there any chance an AI can create original art (not derivative of art created by actual humans) that people will actually find aesthetically pleasing? An experiment at Rutgers University’s Art and Artificial Intelligence Lab sought to find out the answer.
They fed an AI thousands of pieces of art by more than a thousand artists and taught it to recognize different styles. They then had it create its own artwork. They then got experts and members of the general public to view the art alongside works created by people and asked them to judge the art. Surprisingly, not only were humans not able to tell the difference between AI-generated art and that created by humans, they also rated much of the AI artwork higher than the human art.
Of course, just because one AI is able to create attractive artwork doesn’t mean that AI in general can do so consistently. At least not yet.
Take the AI InspiroBot the world’s first (and snarkiest) inspirational quote generator. While the results are often hilarious, they are generally far from actually being inspirational. The same has happened with various other bots in the arts space, including poetry bots, story bots, and music bots. Of course, there are also successful AIs in each of those categories (like the Japanese AI that made it past the first round in a literary contest with a novella manuscript).
Then again, not every AI “artist” is looking to create something entirely unique. Next Rembrandt is a project from Microsoft, ING, and others that looked to create a new Rembrandt painting using AI. The final work of art could definitely have been painted by the master himself.
Apps like Prisma make it possible for your average person to harness the power of AI to have their own photos redrawn in the styles of a variety of famous artists, right on their smartphone.
Critical Reception of AI-Produced Art
While in blind tests it may be difficult for most people to tell the difference between art created by humans and that created by machines, the critical reception of AI art has been overwhelmingly negative.
In an article written for the Guardian, Jonathan Jones attacked Microsoft’s Rembrandt painting as “…a horrible, tasteless, insensitive and soulless travesty of all that is creative in human nature. What a vile product of our strange time when the best brains dedicate themselves to the stupidest challenges, when technology is used for things it should never be used for and everybody feels obliged to applaud the heartless results because we so revere everything digital.”
Then there’s the flipside to the critical reception of AI art: can AI learn to criticize art? The answer to that seems to be, yes. Or at least, maybe. Rutgers University has been carrying out a series of experiments where they teach AI to assess the creativity of various works of art, and so far the experiments have been going well. One portion of the experiment involves telling the AI an incorrect date for a painting to see how it changes the AI’s assessment of the work’s creativity. Ahmed Elgammal had this to say about some of the project’s results:
“We found that paintings from Impressionist, Post-Impressionist, Expressionist, and Cubism movements have significant gain in their creativity scores when moved back to around 1600 AD. In contrast, Neoclassicism paintings did not gain much when moved back to 1600. This makes sense, because Neoclassicism can be considered as revival to Renaissance. On the other hand, paintings from Renaissance and Baroque styles had losses in their creativity scores when moved forward to 1900 AD.”
While assessing creativity is only a small portion of what an art critic does, the fact that an AI still in its infancy can identify originality in works of art based on the time periods in which they are created indicates that AI may eventually be able to delve into more nuanced details of art history and review.
Will AI Push Humans Out of the Art World?
As AI-produced art becomes increasingly better, it is likely that the demand for human-created art may wane. But that doesn’t mean human artists will disappear. Instead, the artists who can figure out how to leverage their humanness will come out on top.
For now, at least, the human brain is significantly more powerful when it comes to creative thinking than AI is. Humans can come up with new styles, new mediums, and new ideas in ways AI isn’t capable of doing (at least not yet). Humans who can then use AI technology to build upon their own creativity to produce new and innovative work will see their own careers flourish.
Inference is one area where AI fails pretty miserably. For example, if I show you a photo of a person holding a cat, and then another photo of the same person with a scratch, you can infer that the cat scratched the person. AI isn’t yet capable of that kind of logical jump. And that’s just one example of an area where the human mind can easily do something that AI just isn’t capable of. Those are the areas where human intelligence and creativity will stay ahead of AI for at least the foreseeable future.
Who Owns the Copyright to an AI’s Work?
One of the more interesting aspects of AI-created artwork is who owns the copyright. Does the programmer who created the AI? The company who owns it? The AI itself? Is it even possible for an AI have intellectual property rights?
In the U.S. at least, copyright law only applies to work created by humans. Therefore, the code that makes up an AI is copyrightable, but the work the AI then produces is not. That is the way courts have repeatedly interpreted the law.
The legal code is a bit ambiguous, though. Nowhere does it state that it only applies to humans even though currently, the U.S. Copyright Office has a policy to reject any applications not submitted by humans.
IP laws were created to protect the interests of humans, specifically to prevent someone from stealing the work of another person in order to benefit from it themselves while depriving the original creator. This doesn’t apply to work created by software, as the software does not benefit from the work it creates.
Eran Kahana, an intellectual-property lawyer at Maslon LLP and a fellow at Stanford Law School, likens it to spell check on a computer. If a software corrects your spelling in a document, does the software then become a joint author? No, and most legal and technology experts would concur that AIs should be prevented from claiming authorship on work they create.
Eventually, intellectual property laws and the way the apply to work created independently by AIs will likely be challenged in court. But until that happens, copyright on AI-created works doesn’t exist (at least in the U.S.).
So What Does This Mean for the Future of Art?
AI artwork is still very much in its infancy, and therefore predicting how soon AI may replace human-created art is difficult, to say the least. But AI technology is moving by leaps and bounds all the time, so it may not be as far off as many artists would like.
Futurist (and current director of engineering at Google) Ray Kurzweil says, “Five to six years ago, AI couldn’t tell the difference between a dog and a cat. Now it can tell the difference between those and can understand thousands of other categories.” What might it be able to do in another five or six years?
The best advice for artists is not to shy away from AI (and other) technology in their work. Those who figure out how to work with AI will have more options than those who take a more ostrich-like approach. And AI can also present opportunities to artists who are paying attention and using their own creativity and resourcefulness to work with the technology.