Wavelength met Todd Yellin who shared insights on how generative AI is affecting the creative production process in entertainment and how the world will need to work together to mediate the ethical considerations presented by AI.
Wavelength met with Todd Yellin, the former head of product and innovation at Netflix who was an integral part of the company’s transformation from a DVD-by-mail service to the global streaming powerhouse we know today.
A film-maker by trade, Todd shares his insights on how generative AI is affecting the creative production process in entertainment and how the world will need to work together to mediate the ethical considerations presented by AI.
A lot of mainstream dialogue about AI is framed in terms of threat. Business leaders, policymakers and the public often ask, “How can we mitigate the dangers – to democracy, to public discourse, to creativity and to our jobs?” Human beings like to think that creativity especially is our secret sauce as a species so it’s unsurprising, to find that Hollywood writers and actors are on the frontline – literally, the picket line – of the debate. Todd is more pragmatic, “People were terrified of the printing press. It starts there, then it becomes the new normal then it becomes more than the new normal. We saw it with television, then video games, it was all scary. We worried about losing what makes us uniquely human, losing our ability to think. Did we overflow the deleterious effects of those new technologies? Probably. Did we underrate the benefits of it? Probably.” The point is, the rapid advancement of this technology is inevitable. The drive for progress, especially in Silicon Valley and the tech centres around the world, is as strong as ever. Even if some countries put specific curbs on it, others will be forging ahead.
Todd’s long career under the bonnet at Netflix is a great reminder of how quickly technological advancements that seem impossible at concept stage and mind blowing at launch stage become … expected. “The idea that you could just click and play a movie – the first time we did it was magical. At first on your laptop, then on a TV. Then we said, how do I skip devices half way through a movie and pick up exactly where I left off? Doing that was really hard, but now it’s taken for granted. Even in the user interface, those scrollable rows like a video store, we made that up.” Rapidly advancing AI is already being absorbed into ordinary people’s everyday lives. We all had a go on Chat GPT and now it’s not just college students who are routinely using it to help with study.
Todd is currently in the writing stage of a movie of his own based in Costa Rica and, with his writing partner, is regularly using Chat GPT to speed up the writing process. “The month I left Netflix was the month Chat GPT 3.0 was released to the public. Now I’m trying to create a fictional world but I need to capture an authentic feel. I’ve done my research, I’ve been on the ground but when I’m writing about seedy bars in downtown San Jose in 2010, Chat GPT is much faster than Google to find out about them.” Yellin is adamant that LLMs can’t create whole scripts, at least not good ones. “When I ask it to write a scene it turns out crap but occasionally there’s a nugget in there.” The point is that the LLM can’t think of the ideas, it can only build on ideas that are already in there and feed into the system. “It’s helpful, it’s a tool. Why deny yourself the use of it?”
The technology has unlocked other possibilities that are already in use across the entertainment industry, and Yellin is realistic about both the pros and the cons. “What streaming sparked and what Netflix really capitalised on was global release of content, all at the same time. It doesn’t matter what language it’s in, you could get access to that.” Watching a foreign language movie in the States used to be a niche pass time but through shows like Narcos and Squid Game, Netflix has started breaking down some of those cultural barriers. AI tech can take this one step further. “Nobody wants to watch badly dubbed movies and if you’re reading subtitles you’re missing some of the art of what’s going on – so what if we can start dubbing into English in an artful way, by matching the actor’s lips and using their voice? Netflix is experimenting with that.” It raises challenging questions for the acting unions as viewers will be watching a performance by an actor that, strictly speaking, they didn’t deliver. Yellin says, “Entertainment is one big magic show. If Emma Stone’s acting can be shared with more people who don’t speak English, provided people are compensated fairly and we use the technology with our eyes wide open, that is the idea.”
AI is also being adapted to speed up the process of getting a movie into production in the first place. Yellin explains how an image creation tool called Midjourney is being used by film makers to visualise a movie before they even start shooting it, featuring the images of the actors they want to cast, the world they want to create. “Getting people to read is a pain in the ass.” says Yellin, “with Midjourney, you’re practically watching the movie in stills which helps in production design, costume, shot creation with cinematography…”
Yellin was a well-regarded documentarian before he worked for Netflix, but the rules about documentary filmmaking and news reportage are not the same as with, say, making a fantasy movie, but nor are they the same as each other. “Once you start introducing AI deep fakes into news, it’s really scary. But with documentaries, they’ve always included dramatic recreations or representations to help tell the story. Every editing decision is a point of view.” Viewers should expect that documentaries will include AI generated footage – the question is how and whether to indicate what is and isn’t “real”, which of course raises its own set of ethical questions beyond fair remuneration.
Who should be responsible for mediating what is and isn’t acceptable? “In an ideal world it would be our elected officials” said Yellin, recalling the incident in 2018 when Mark Zuckerberg appeared at the senate and was asked to explain the business model of Facebook. “We need more informed representatives and officials. Responsible entrepreneurs and founders, backed by demanding investors, are part of the mix. As are the media who can help to educate the public.”
Yellin’s advice to CEOs lying awake at night worrying about what all this could mean is simple. “Don’t take a New York Post (i.e. tabloid newspaper) style headline approach to your attitude to AI.” In other words, keep your eyes wide open and find out everything you can from people who know what they’re talking about. Yellin recommends reading God Human Animal Machine by Meghan O’Gieblyn. It’s a “strikingly original exploration of what it might mean to be authentically human in the age of artificial intelligence.” Sounds like a good place to start.