Of all the dimensions of Elon Musk that fascinate his fans—his intellect, his work ethic, his rockets, his dating life—there’s one that hasn’t been definitively explained: his seemingly self-restoring hairline. In early career moments, like a 1999 CNN segment that followed Musk getting a McLaren F1, or a 2000 image of he and Peter Thiel hyping PayPal, normal signs of male pattern baldness are visible. Since then, the CEO of SpaceX and Tesla appears to have found a solution, but it’s his earlier incarnation that an experimental artist has plastered across San Francisco this week.
And that artist has a theory. “I’m pretty blown away at how good his hair plugs are,” says Katsu, whose mysterious posters appeared in San Francisco and New York City. “It’s like the perfect metaphor of what you get when you reach his level of success. You get your fucking hair back.”
The posters are not a personal attack though: They’re promoting “A.I. Criminals,” Katsu’s new art show opening this weekend in San Francisco, which explores the idea of artificial intelligence as this generation’s atomic energy, where the race to profits outweighs the risks. And although Katsu has famously taken aim at billionaire technocrats before—like his 2015 portrait of Mark Zuckerberg painted with excrement or a 2013 poster of Zuck with a black eye and chipped tooth—Katsu frames Musk as a Cassandra, not the titular criminal. The show explores the risks of artificial intelligence, so Musk—a crusader for the same cause—seemed like a good fit, Katsu says: “We’re going into an age where titans of tech are mutating into these weird idols in our head.”
The exhibit’s centerpiece is a series of “AI portraits” that delve into the use of AI in profiling and law enforcement and the way that these new technologies “begin to see us,” he says. The show will be open to the public on Saturday and Sunday at 128 Texas Street in San Francisco.
For the portraiture series, Katsu used Google’s Tensor Flow machine-learning library to develop a type of AI system called a Deep Convolutional Generative Adversarial Network. The system was designed to generate unique portraits based on what it learned and trained using a data set of vintage black and white mug shots that Katsu scraped from the web. Training models will spit out progress reports to help tune these systems, he says, and those outputs will be the portraits: “Some of the faces are scrambled and repurposed … You just start to see this kind of system of how they look at eyes or lips or noses. It’s a little haunting and ghoulish but also really beautiful.”
Katsu has reappropriated technology in unintended ways before. In 2015 he hacked a Phantom drone to vandalize a billboard of Kendall Jenner (and jump-start the idea of drone graffiti). For “A.I. Criminals” he got some help from F.A.T. Lab, a community of creative technologists and hacker artists. (He previously collaborated with the group on a graffiti app.) Katu says he takes pride in demonstrating that new technologies are not solely the province of engineers: “They’re not reserved for the dev community only. Github is actually insanely easy to use if you try. There’s certainly a Valley culture—and maybe it really came out in the Google memo—[of being] exclusive and fear of sharing this skill set.”
There’s more to the AI portraits, in Katsu’s view, than just a comment on the growing power of technology—it’s a warning about humanity’s lack of control over what these tools can build. “We live in a time where an artist with the right video-graphics cards can begin to teach a computer to learn certain things,” he says. “Imagine malware generated from AI systems that were trained on deception.”
It may not even matter if the creator’s intent is malicious. By building systems trained on “biased data generated from biased policies,” Katsu says, we will perpetuate the cycle: “We’re teaching our offspring our same biases. It’s like a racist father unintentionally making a racist son, or my dad turning me into a Democrat.”
Ultimately, then, “A.I. Criminals” is an indictment not of artificial intelligence but of the people blithely pushing the technology past the point of prediction. “Maybe I play too many videogames and watch too much anime,” he says, but the possibility of the Singularity should be a more imminent concern: “In no time at all we will be interfacing with a new type of entity. We make a lot of assumptions of what this will be like. It’s really egotistical of us.” Musk, whom Katsu compares to Blade Runner‘s Rick Deckard, hunting down artificially intelligent replicants, is on the same wavelength.
If you're not concerned about AI safety, you should be. Vastly more risk than North Korea. pic.twitter.com/2z0tiid0lc
— Elon Musk (@elonmusk) August 12, 2017