Justine Bateman won’t name names, but a TV showrunner friend once came to her with a dilemma: their show’s team was well into filming its second season when a network executive had an idea. A character in the pilot hadn’t tested well with audiences, so they were just going to go in, use a little AI, and swap in someone else.
The showrunner – and Bateman, an actor and director – were understandably incensed. “When you change the beginning of something, you change the creative trajectory,” says Bateman. “There’s going to be whiplash for the viewer when they get to episode three or four because what was set up in the pilot got messed with and now doesn’t make sense.” Using AI might have seemed like a simple solution to the executive, but to the showrunner, it was catastrophic.
But AI is creeping into television more and more – and it could come at a major cost to the industry. Though the WGA and Sag-Aftra made a lot of noise about AI’s job-stealing potential in last year’s Hollywood strikes, a recent report by CVL Economics says it’s still likely that 203,800 American entertainment jobs will be “disrupted” by AI by 2026. This makes the technology a frightening agent of change for television, and the entertainment industry at large.
While the use of AI is still regarded as a sort of landmine in Hollywood – with most tiptoeing around public scepticism and the (justified) perception that they’re replacing humans with computers – that doesn’t mean it’s not everywhere already. “A lot of what people have billed as ‘AI’ is stuff that was in use over the last decade,” says Emily St James, a TV writer, podcaster and cultural critic. Disney+ used Respeecher to turn the weathered voice of then 68-year-old Mark Hamill into that of a 20-year-old Luke Skywalker for The Mandalorian.
The BBC and Amazon MGM Studios have admitted to using AI to make marketing materials for shows such as Doctor Who and Fallout (including a bizarrely inaccurate AI image of Los Angeles). And Banijay, the global conglomerate behind more than 200 reality shows, including Deal or No Deal and Keeping Up With the Kardashians, just launched an AI Creative Fund to make new programmes. It’s already released Fake Show, an Italian series in which celebrities have to improvise comic scenes generated by an AI.
That doesn’t mean AI can create a whole show – yet. “There’s still a significant gap between machine-learning simplifying technical processes in ways a human could not and ‘taking over television’,” says St James. “Maybe I am whistling past the graveyard, but I have yet to see anything more convincing than hype.”
Others agree, with creative director Lauren Fisher saying: “Everything I’ve seen that is purely ‘let the AI do the whole thing’ turns out to be shit, but because it’s a novelty, it gets oohs and aahs.” TV producer Benjamin Field adds: “AI doesn’t do quality. AI does just about good enough.” As one showrunner quips anonymously: “AI will never replace good work for the same reasons sex robots will never replace the real thing.”
“I think AI can do the grunt work,” says Guy Branum, a writer and producer who’s worked on shows such as Hacks, The Other Two and The Mindy Project. “I once had a job writing introductions for the VMAs. That job was basically copying old introductions and updating them to include the current talent. ChatGPT, with proper supervision, could do that job. But all it can really do is steal from existing works and recapitulate them in the most formulaic way.” If a show has an established blueprint, though – like, say, Big Brother or RuPaul’s Drag Race – then it could be more a question of “when” rather than “if” we’ll see AI-generated challenges and scripts.
Because it’s so reflexive, most experts say AI probably won’t ever write great comedy, or at least topical satire like you might see on shows such as Last Week Tonight With John Oliver. Still, there’s a lot of TV that’s not sharp or timely, from Netflix dating shows to Hallmark Christmas movies, all of which you can imagine being influenced by AI material. Of course, with more channels and streaming services comes more content, all of which has to be written by someone – or something. “Over the next six to 12 months, we’ll see FAST (free, ad-supported TV) or social media channels developing AI content” Field says. “It’ll probably be made by a minimal amount of creatives and there’ll be no real money available for it, but it’ll exist out there in the ether. It may do some interesting things, but it isn’t going to change the world. It’s just done cheaper and worse.”
That doesn’t mean creators aren’t begrudgingly embracing this technology. Field himself has co-founded a company dedicated to making “ethical, policy-driven” synthetic media. “We’re in a perfect storm where budgets are reducing and AI has cthe ability to create content at a cheaper rate and faster,” says Field. “I don’t necessarily see it as something that’s going to railroad us and steal all the jobs, but I do think we as an industry have to behave better and work more sustainably toward the future … taking every tool at our disposal and making something new with it.”
Other creatives are working to lay sandbags around the rising tide of AI. The National Association of Voice Actors (Nava) has taken a proactive stance over what it says are the abuses being levied against its group by overzealous executives and producers. They are working not only to change the voiceover industry but also lobbying US congress for legal protection through the No Fakes Act and the No AI Fraud Act, both of which are now making their way through the legislative process.
Unlike a person’s name, image and likeness, copyright law doesn’t currently cover a person’s voice. That’s especially alarming in 2024, when just three seconds of someone’s “voiceprint” can create a believable copy using AI. (In 2020, you needed about six hours of audio to achieve the same result.) That has resulted in some high-profile upsets: Scarlett Johansson was “angered” when she found out ChatGPT had used a voice without her consent which “sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference.” There have also been some startling deepfakes, including one that used the voice of a faux-President Biden to discourage people from voting in a state primary election.
In the world of entertainment, these cheap and easy voiceprints have made it far quicker to try to grab a busy actor to re-record a few lines, but they’ve also allowed companies to create wholesale dialogue using actors’ existing work. Cissy Jones, a veteran Bafta-winning voice actor, says that a few years ago, she found her voice in multiple works on multiple websites, despite having never contributed to the projects. Fans of The Owl House – a show she appeared on – were making edits of videos using her voice, but saying lines she’d never recorded and that, she says, “turn pretty pornographic fairly quickly.”
“It was done without my consent, certainly without any control to how they were used, and no compensation whatsoever,” Jones says, adding that while she’s an adult, she’s especially troubled when these AI works use the voice of child actors, so “you can imagine how dark that gets real quickly.”
Jones’s voice had been fed into an AI engine – something the actor says she’s heard sound engineers have been asked to do by clients to avoid paying for script changes. Nava wants voice actors to be able to decide whether they’d like a digital replica of their voice, as well as control over, and compensation for, any time their voice is used. To ease this along, the group has created Ethovox, its own database of actors’ voiceprints, all of which can be used by creators.
after newsletter promotion
While the fan edits featuring Jones’s voice were unsanctioned, some are optimistic that AI might create alternate-universe versions of fans’ favourite shows. While the idea that studios could just knock up new episodes of I Love Lucy or The Brady Bunch might seem jarring from an artistic standpoint, some are banking on animated series like The Simpsons or South Park seeing the value (and earnings potential) in allowing fans to make their own “episodes”.
Edward Saatchi, CEO of Fable Simulation, says his company Showrunner is trying to put TV production into the audience’s hands. “Maybe you’re at home and you finished a particular season of a television show,” he says. “You click to get the next episode, and maybe you say, ‘put me in that episode. I want to be friends with the hero,’ or ‘here’s roughly what I want to happen.’ We want to make it easier to tell original stories and to make a TV show about your life.”
Showrunner is also making its own shows, such as Exit Valley, an animated satire “starring” Silicon Valley billionaires. It’s already released two episodes and has invited its 1,000 users (there are 7,500 on the waiting list) to make even more. They can cobble these together by inputting 10 to 15 word prompts, yielding fully fledged scenes from two to 16 minutes long. A jury of film-makers and creatives has been selected to pick the 20 best episodes, and Saatchi says episode creators will get a cash prize and a share of the revenue from streams.
“We want people to create something that stands the test of time,” says Saatchi. “You have to plausibly think that you could sit down with your friends and love watching that episode of Exit Valley. You have to believe it’s actually cool, not just because it’s made with AI.”
So-called amateur creators dominate the market elsewhere, too, with 295 million people – slightly more subscribers than Netflix has worldwide – tuning in to Mr Beast’s YouTube channel every week to watch the site’s most popular star unveil his latest outlandish stunt. In fact, YouTube is already the most popular streaming service on television, drawing about 10% of viewership on connected TVs in the US, more than any other service, as well as an estimated 4.95bn active users worldwide.
“Traditional television is already being disrupted,” says Doug Shapiro, cultural commentator and media expert. “For years, the argument has been that YouTube is not competitive, that it’s not professionally producing content. But that 10% doesn’t reflect viewing on mobile or PC. It’s only from people sitting down in their living room, turning on their TV and watching YouTube. That disruption from the bottom is already happening. The real question now with AI is whether these tools throw gas on the fire, so the tens of millions of creators already making their own stuff continue to move up the quality and performance curve, becoming increasingly competitive with Hollywood for people’s time.”
“To use a Game of Thrones analogy,” adds Shapiro, “Netflix v Disney is like the Targaryens v the Lannisters – but the army of the dead is at the wall, and that’s the tens of millions of individual creators.”
When professionally made and consumer- or AI-made content start becoming interchangeable, the sea change could have major implications for Hollywood beyond those 200,000 disrupted jobs. If DIY AI TV takes off, it will turn the entire notion of television on its head, upending art and fame as we know them. Or as Saatchi puts it, if AI continues to make inroads into television, “you won’t get 15 minutes of fame. You’ll get seven seasons and a syndication deal.”