Ume is now back and on a mission – to commercialize video deepfakes for the planned metaverse and make them as central to digital life as tweets and memes.
He will take that next step Tuesday when a deepfake developed by Metaphysic, the company he formed with entrepreneur Tom Graham, will compete in the semifinals of the NBC reality hit “America’s Got Talent.”
“This is a good opportunity to raise awareness and show what we can do,” Ume said.
“We think the web would be much better if we lived in the world of the hyperreal instead of avatars,” Graham added, describing users’ ability to manipulate real faces with Metaphysic.
The start-up’s appearance in front of millions on TV will lay the foundation for its new website that should make it easier for ordinary people to say and do things they never did in real life. Many other such sites are aimed at programmers and researchers.
And the deed – in which they will follow a raucous performance in the preliminary round in which they placed the face of a young Simon Cowell on the screen above a stage performer so that the judge appeared to be singing to himself – offering a shiny advertisement for a technology that is democratizing with astonishing speed.
Still, some critics are shocked by this celebratory moment in a top-rated television show. Video deepfakes, they say, blur a line between fiction and reality that is barely clear now. If disinformation peddlers can have so much success with words and edited images, imagine what they could do with a full-length video.
“We’re quickly entering a world where anything, even videos, can be manipulated by just about anyone who wants to,” said Hany Farid, a professor at the University of California at Berkeley and an expert on deepfakes. “What can go wrong?”
The reveal of what has been the most-watched program on network television for most weeks this summer comes at the end of a hectic summer in the world of deepfakes, who use the deep learning of artificial intelligence to create fake media (supporters prefer ” synthetic” or “AI-generated”).
While many Americans indulged in scenic analog activities like going to the beach, a start-up called Midjourney offered “AI art-generation,” where anyone could create stunningly real images with a simple graphics card with a few keystrokes. To spend even a few minutes with it – there Gordon Ramsay is burning in his Hell’s Kitchen; here’s Gandalf shredding on a guitar – is to experience a technology that makes Photoshop look like Wite-Out. Midjourney has amassed over a million users on its Discord channel.
Three weeks ago, a start-up called Stable AI released a program called Stable Diffusion. The AI image generator is an open-source program that, unlike some rivals, places few restrictions on the images humans can create, leading critics to say it can be used for scams, political disinformation, and privacy violations.
“We should be concerned. I follow the technology every day, and I’m concerned,” said Subbarao Kambhampati, a professor at Arizona State University’s School of Computing & AI who has studied deepfakes and virtual identities. He said he expects the “AGT” moment to see platforms like this rise even further, even as the technology improves day by day.
“It goes so quickly. Soon everyone can [to] make a moon landing that looks like the real thing,” he said.
Ume and Graham say that cheating is not their goal. Ume emphasizes the entertainment value: The company will market to Hollywood studios that want to present deceased actors in movies (with estate permission) or have performers play against their younger selves.
As for regular users, Ume says the goal of Metaphysic is to make online interactions feel more real — none of the quirkiness of video games or the flatness of Zoom. “I imagine I can have breakfast with my grandparents in Belgium from here in Bangkok and feel like I’m really there,” said Ume from his current base.
Graham adds that synthetic media will not harm privacy, but will enhance it. “I would like to see a world where online communication is a more humane experience owned and controlled by people,” said Graham, a Harvard-educated attorney who founded a digital graphics company before turning to crypto and eventually deepfakes. . “I don’t think that happens in the Web2 world today.”
Farid is not convinced. “They only tell half the story—yours using your own image,” he said. “The other side is that someone else is using it to defraud, spread disinformation and disrupt society. And you have to ask yourself whether it is worth moving around a bit more on Zoom.”
Deepfake technology started using “generative hostile networks” eight years ago. Created by computer scientist Ian Goodfellow, it essentially pits two AIs against each other to compete for the most realistic graphics. The results were much better than basic machine learning techniques. Goodfellow would go on to work for Google, Apple and now DeepMind, a subsidiary of Google.
In the beginning, deepfakes were used by skilled exploiters, who infamously graft the faces of actresses onto pornographic videos. But with the technology requiring fewer tools, it can now be harnessed by everyday people for a range of applications, which Metaphysic hopes to advance.
The company raised a $7.5 million investment earlier this year from the likes of the Winklevoss twins, the social media turned crypto entrepreneurs, and Section 32, the VC fund of original Google Ventures founder Bill Maris. “We believe the impact will be far-reaching,” Andy Harrison, managing partner at Section 32, said of Metaphysic. Harrison, also a Google veteran, said he saw video deepfakes not as a threat, but as an enlivening change in consumption and communication.
“Honestly, I’m pretty excited,” he said. “I think it’s a new era in entertainment and social interaction.”
Critics, however, are concerned about the “liar’s dividend,” in which a web awash with video deepfakes is clouding the water even for legitimate videos, leaving no one to believe anything.
“Video has been the last frontier of verification online. And now it may also be gone,” Farid said. He cited the unifying power of the George Floyd video in 2020 as improbable in a world overrun with deepfake videos.
Asked about “AGT”‘s role in promoting deepfakes, a spokesperson for production company Fremantle declined to comment on this story. But one person close to the show who asked for anonymity because they were banned by law from commenting on an ongoing competition said they believed the metaphysics act had social utility. “By using the innovation in a completely transparent way,” the person said, “they are showing a mainstream audience how this technology can work.”
A solution to the truth problem could come in the form of authentication. A cross-industry effort involving Adobe, Microsoft and Intel would verify the creator of each video and make it transparent to assure people it was real. But it’s not clear how many people would adopt it.
Kambhampati, the ASU researcher, said he fears the world will end in one of two places: “Either nobody trusts what they watch anymore, or we need a comprehensive authentication system so they do.”
“I hope it’s the second,” he said, then added, “not that it seems so great either.”
0 Comments