Natasha Lyonne On ‘Uncanny Valley’s “Copyright-Clean AI” Use

EXCLUSIVE: Part of Hollywood’s cohort of ethical AI advocates, Natasha Lyonne is walking the talk with her upcoming feature directorial debut.

Ahead of the Season 2 premiere of Poker Face, returning May 8 with three episodes on Peacock, the 2x Golden Globe nominee told Deadline how she plans to utilize artificial intelligence in Uncanny Valley, in contrast from the “dirty model” on which the tech is often built.

“I just heard we have this first clean foundational model that is copyright-clean that you can build on, but ultimately that is a tool,” explained Lyonne. “That’s not going to omit any department heads or production designers or cinematographers.

“This movie I’m doing with Brit Marling is actually a real film. It’s almost, you could think of it as green screen or something like that, just for now, so you’d be able to do a film at a greater scale, which is exciting for filmmakers and also production designers and the like,” she added.

Last month, Lyonne joined Paul McCartney, Ava DuVernay, Taika Waititi, Cate Blanchett, Alfonso Cuarón, Lilly Wachowski, Ben Stiller and more than 400 entertainment industry professionals in urging the Trump administration to strengthen copyright rules against AI amid pressure from tech companies.

“Everybody should be heavily alarmed,” said Lyonne. “Ring that alarm, folks. But in the short term, I think it’s important to preserve what we’ve got, just for today.”

The Poker Face star emphasized the “real problem” of data theft and fair use currently at play with tools like OpenAI and ChatGPT, as opposed to regulated forms of AI like the program she plans to use for Uncanny Valley.

“I do think that it’d probably be wise as a community to start distinguishing the details here, because if we’re just jumping at every AI turn, then we should all be throwing our cell phones away,” she said. “I do think that Adam McKay recently pitched that, which was a full shutdown because of the climate issue that we’re talking about is again unquestionable.”

Lyonne continued, “I think that right now, we’re really up against it. There’s a lot of talk about within the next five years, this kind of AGI [artificial general intelligence] and all this sort of stuff that I think we really need to be paying attention to rather than just having a trigger finger that creates an inability to parse information correctly. I think that lack of nuance in the modern world is a real danger if we want to get anything done about anything.”