If a well-trained AI can capture the essence of a subject, as Sam Altman has argued, then an LLM trained on Altman should basically be him, right?
That’s the turn-the-tables premise of Deepfaking Sam Altman, a new documentary produced and financed by Kevin Hart’s Hartbeat shingle based on a well-received New York Magazine story about the OpenAI leader. (“Sam Altman Is The Oppenheimer Of Our Age,” was the 2023 piece’s headline.)
Riding a buzzy screening at sxsw in March, Deepfaking Sam Altman will now get an imminent release from Abramorama, the boutique theatrical distributor that has worked on acclaimed docs like The Last Class, Senna, Anvil! and Exit Through the Gift Shop, The Hollywood Reporter has learned. The movie will open in New York on January 16 with a national rollout to follow. Abramorama CEO Karol Martesko-Fenster said in a statement that the film is “exactly the kind of bold, culturally relevant documentary that deserves a theatrical experience.” Vox Media Studios, sister company to New York Magazine, also produced.
Directed and produced by Adam Bhala Lough, who was behind HBO’s well-reviewed docuseries Telemarketers in 2023, Deepfaking began when Lough optioned the magazine piece, then sought an interview with Sam Altman. Hearing crickets, he showed up Michael Moore-ishly to the San Francisco headquarters of OpenAI only to be promptly removed.
So Lough decided on a different path: he would train an LLM on everything Altman ever said or wrote, create a deepfake of the entrepreneur; then graft the image onto an actor and interview him. If Altman was right, this should be Altman.
This Andy Kaufman-esque plan ran aground, however, when no U.S. company would deepfake his subject. “I think people are really afraid of Sam Altman,” Lough said in an interview. “Every time I’d bring up his name in Los Angeles or San Francisco, people would get scared.”
The Punjabi-American director decided to go to India, where he has family, and found a company to do it there, where there is presumably less Altman trepidation. The result is the film, which in creating a conversation with AI Altman (he calls him “Sam Bot”) seeks to interrogate the quickly vanishing difference between human and machine intelligence, not to mention the fungible nature of identity when machines will represent us. Lough’s interview with Sam Bot showed the ways AI can capture at least the externalities of our personality if not more.
Maybe even more instructive, Lough says, is the ways he learned that we need to be careful around it. “My biggest takeaway is that we need to think about AI as if we were raising a toddler,” the director told THR. “As any parent knows you need to be very careful what you say around a toddler because it’s soaking up everything you’re saying. AI is not a adult and it’s not going to take over the world but that doesn’t mean it isn’t extremely powerful.”
Hart is fascinated with AI himself and initiated contact with Lough after he watched Telemarketers (which in satirically uncovering scandal at a telemarketing sweatshop dabbles in its own themes of techno-dislocation). Hartbeat this fall partnered with Hollywood AI company Luma for a “Live AI Battle” to create a machine-driven film in real time. In a statement, Luke Kelly-Clyne, who heads the studio at Hartbeat, said Lough’s movie “captures the curiosity we’re all feeling about AI and what it means for us, but does so by turning it into a smart, funny, and surprisingly human story.”
In addition to Amazon MGM’s upcoming film about Altman (Luca Guadagnino’s Artificial stars Andrew Garfield in the lead role), Hollywood is reckoning with Generative AI more broadly. Studios by turns are suing AI companies and looking to bring the tech onto their platforms, while some creators lustily embrace it and others declare in-show they won’t use it.
Lough says that he was not trying to issue a broadside against the tech. “I’ve already had some people ask why I didn’t slap it down more. I get this is a polarizing issue, but I also don’t think it’s so black and white.”
So far, at least, Lough has not heard from Altman or his lawyers. If he does, one can imagine a ChatGPT response. OpenAI’s LLMs have some pretty good legal training.
