“If you've seen this video, you probably don't want to admit it,” I told the audience. “It's the face of Wonder Woman, Gal Gadot, superimposed onto the body of a porn star and uploaded to a website called ‘Step-siblings Caught,’ purporting to show her having sex with her stepbrother.”
I told them about the ability to put your face on someone else's body. To use your face as a remote control for someone else's face. To generate convincing audio in someone else's voice. I showed the video of Jordan Peele making Obama say “Stay woke, bitches” and got a good laugh.
I shared these things because I wanted people to become aware of how the technology is evolving. I wanted people to be more judicious about what they believed and what they chose to share. I wanted people to become more critical of what they saw and heard online.
In the time since, what is commonly referred to as artificial intelligence has, of course, evolved exponentially. And this particular brand of artificial intelligence -- the kind used to generate convincing video and audio of things that never happened -- has really come into its own.
Whether it's a video of Tom Cruise tripping over his own feet or Anthony Bourdain posthumously narrating his own life, convincing deepfakes are now widespread.
But perhaps it's not the celebrity impersonations we need to be worried about. Perhaps the real tragedy of this form of AI is that it undermines our trust in any content.
In a recent conversation with Kelly Catalfamo at Futurism, Simon DeDeo, an assistant professor at Carnegie Mellon University and external faculty at the Santa Fe Institute, said his current nightmare is AI-generated "people" driving actual vulnerable humans literally crazy. And that doesn't require nearly as much technological sophistication.
"I can con you without needing to fake a video," he said. "And the way I con you is not by tricking your visual system, which the video deepfakes do. I con you by tricking your rational system. I con you at a much higher level in your social cognition."
Part of the problem is that we have already created a fake society. DeDeo talks about QAnon being “sustained by algorithms that fake sociality, that fake friendship, that fake prestige, that fake all of these things that are just kind of basic to our cognition” -- but doesn't that description fit our entire online life?
After all, the fake people are becoming realer and the real people are becoming faker, and we are all just going to meet in the middle and wonder how we got there.
At a minimum, AI is bringing us face to face with our own mediocrity. “‘There is a lot of not-thinking that human beings do,’ said DeDeo, who, as a professor, has read his share of formulaic essays. ‘There is a lot of things people say that sound smart but actually have zero content. There is a lot of articles that people write that are meaningless. [AI] can imitate those to perfection.’”
Maybe we should be less concerned about how smart artificial intelligence is -- and more worried that we ourselves were actually never that smart to begin with.
The con has existed forever, we didn't need AI/deepfakes (or even the internet) to create it. Will they exacerbate it? Absolutely.