Artificial Intelligence

AI Is Coming for Hollywood's Jobs

But so is everyone else.

|

Joanna Andreasson/DALL-E4

Tom Cruise might just be Hollywood's most analog movie star. He reportedly once grew irate when a crew member suggested that a dangerous stunt be performed by a digital double, yelling: "There is no digital Tom! Just Tom!"

For last summer's Mission: Impossible—Dead Reckoning Part One, Cruise, who was 61 when the movie hit theaters, actually jumped a motorcycle off a ramp on top of a mountain, let the bike fall down into the canyon, and then parachuted down into the valley below. The complex sequence took a year to plan and shoot. It probably cost a lot of money. There were some computer effects involved, including digitally converting a bike ramp into a stretch of rocky mountain. But there was no digital Tom.

Like many of Hollywood's top-tier talents, Cruise has spent much of his career fighting against digital encroachments into Hollywood's processes, especially those that might replace real people. So it was no surprise that Dead Reckoning pitted Cruise's longtime franchise superspy Ethan Hunt against an omnipresent artificial intelligence known only as "the Entity," an enemy that was described in the film as "everywhere" and "nowhere," capable of accessing any digital system and, in the process, "compromising the very truth as we know it." It was a fitting metaphor for Hollywood's own fearful struggles against AI.

As the movie rolled out in theaters, Hollywood's actors and writers unions were going on strike. The unions were concerned about the usual issues—pay rates, benefits, contract transparency, and work expectations. But as much as anything else, they feared for their jobs, worrying they would be made obsolete by generative AI.

"We want to be able to scan a background performer's image, pay them for a half a day's labor, and then use an individual's likeness for any purpose forever without their consent," is how the Screen Actors Guild-American Federation of Television and Radio Artists, the actor's guild, characterized the Producers Guild of America's position. "We also want to be able to make changes to principal performers' dialogue, and even create new scenes, without informed consent. And we want to be able to use someone's images, likenesses, and performances to train new generative AI systems without consent or compensation." The Writers Guild of America, meanwhile, worried that screenwriters might be forced to write drafts based on AI outlines, respond to AI notes, or find their own original work rewritten and restructured by AI software.

No one, in other words, wanted to work for a robot. No one wanted to be replaced by one. The industry was united in defiance of digital Toms.

But work with a robot? That was a slightly different story. As it turns out, AI tools, even in their infancy, are already quite useful assistants, especially for big-picture organizational tasks like outlining. Crafting the arc of a story in a Hollywood script can be as big a part of the task of producing a screenplay as writing the specific lines of dialogue that make up the scenes; screenplay structure is a complex art unto itself.

So when the Writers Guild cut a deal to end the strike, its contract included provisions allowing writers to use AI to assist with their own work, though no AI could be credited. Before the robots revolted, they would be partners, making work easier and more efficient for creatives.

You can tell a lot about an industry from what it's afraid of. For decades, Tinseltown entertainers have conjured up lurid tales about the dangers of artificial intelligence. From 2001: A Space Odyssey to Westworld to The Terminator to The Matrix, intelligent machines have been typecast as villains—inhuman and uncaring, relentless and powerful, bringers of death and destruction. Hollywood's AI antagonists are murderous deities and gods of technological apocalypse. They are not helpers or tools, but powerful alien minds indifferent to ordinary human suffering—not, one suspects, unlike some movie studio executives.

Now those fears have migrated into the real world. Both strikes occurred in the aftermath of high-profile advances in generative AI, with programs such as Midjourney showing their power to produce impressively vivid and sometimes lifelike images based on simple text prompts, and AI chatbots such as ChatGPT demonstrating that AI could produce high-quality written products such as essays.

Writers and actors both feared they would be forced to work for the AI tools—or worse, that they would be replaced entirely. A movie star such as Cruise could insist there would be no digital Tom, but rank-and-file Tinseltown talent were worried that a real-life Entity might come for their jobs and no Hollywood hero would save them.

In many ways, they were—and are—right to be worried: AI threatens to upend just about every aspect of cinematic production, not just writing and acting. Animation, special effects work, makeup, costuming, lighting, photography, and set and production design are all likely headed for some form of disruption as generative AI tools make it possible for ordinary people without specific training to conjure up and manipulate high-quality audio and video.

But therein lies the opportunity. AI might wreak havoc on traditional studio moviemaking, with its massive budgets and complex technical requirements. But in the process, it is likely to make high-quality filmmaking much less expensive and logistically arduous, empowering smaller, nimbler, and less conventional productions made by outsiders with few or no connections to the studio system.

'I Can Sit in an Office and Do This With a Computer'

Making a movie is an absurdly, ridiculously, almost comically expensive endeavor. Blockbusters regularly cost $200 million or even $300 million to produce, and marketing costs for big films can reach or exceed that figure. Dead Reckoning, for example, cost a reported $291 million to make before a single advertising dollar was spent.

Smaller productions are pricey as well: A modestly budgeted feature might still cost millions to produce. Some episodes of television can run past the $20 million mark, even without Cruise and his motorcycles.

The high cost of production makes the film business extremely, almost irresponsibly, risky for both creative talent and film industry executives. This affects the industry's output: The need for many big-budget films to earn well over half a billion dollars at the global box office has narrowed much of Hollywood's output down to a few surefire genres and categories, such as animated family films and superhero movies.

Where does all that money go? To pay people. Watch the final credits on even the smallest movie, and you are likely to see dozens of names. The biggest productions are vast enterprises, with hundreds upon hundreds of artists, technicians, and skilled craftspeople involved. Every single name that scrolls by represents a paycheck added to the budget.

One of the largest cost categories for modern films and movies is computer-generated effects work. Individual episode budgets for She-Hulk: Attorney at Law, the Marvel TV show for Disney+, ballooned wildly during production, hitting $25 million apiece—more than the finale of HBO's epic fantasy series Game of Thrones—in large part due to the cost of the digital effects used to create the main character, who is bright green and nearly 7 feet tall. It was a big character, a big expense, and, partly as a result of the cost, a big failure.

But subtler special effects are now part of many prosaic productions that are not obviously special effects showcases. Performers in more realistic productions now regularly appear de-aged using subtle but expensive digital effects. Virtual production, in which scenes are shot on sets built out of walls of high-resolution LED panels, allows filmmakers to shoot scenes in almost any environment without constructing physical sets or traveling to film on location. This sort of shooting gives filmmakers flexibility, but the facilities are expensive to build and to staff.

What if the cost of virtual production were to drop dramatically? What if sets and locations and backgrounds could be generated easily, at home, with desktop computers using simple text prompts? And what if, at the same time, digital effects and makeup work, like that used to create She-Hulk or de-age older actors, became just as accessible?

That is exactly what the coming generation of AI filmmaking tools promise.

In November 2023, Jeffrey Katzenberg, the longtime studio executive who presided over one of Disney's most artistically and commercially successful runs of animated features in the 1990s, predicted that AI would reshape every aspect of the industry, especially animated features. "I don't know of an industry that will be more impacted than any aspect of media, entertainment, and creation," Katzenberg said at the Bloomberg New Economy Forum. "In the good old days, you might need 500 artists and years to make a world-class animated movie. I don't think it will take 10 percent of that three years from now."

In Katzenberg's telling, it wasn't just that a $200 million Pixar film could become a $20 million movie. The centrality of animation and effects work to contemporary filmmaking meant that a far greater number of productions could be affected. Instead of an overbudget mess, the next She-Hulk might be made on a comparative shoestring.

Just a few months after Katzenberg's prediction, OpenAI, the company behind ChatGPT and arguably the leader in generative AI technology, announced Sora, a program that could generate incredibly vivid computer animation from simple text prompts. Image quality varied, with some scenes looking practically photoreal and some looking decidedly more clunky. But by mid-March, Bloomberg reported, OpenAI was meeting with Hollywood players about using the software, and some filmmakers were experimenting with its capabilities.

This was just the initial release. Ensuing generations would almost certainly become more and more advanced. If nothing else, many of the effects looked better than the expensive, cruddy effects found in She-Hulk.

The writing was on the LED wall.The old, time–onsuming, expensive methods of film production—whether virtual or physical—were on the way out.

Days after the debut of Sora, Tyler Perry, the actor, writer, director, and independent film mogul who today owns a large production studio in Atlanta, announced that he was canceling an $800 million studio build-out, citing what he'd seen from Sora.

"I had gotten word over the last year or so that this was coming, but I had no idea until I saw recently the demonstrations of what it's able to do," Perry told The Hollywood Reporter. "It's shocking to me."

What, exactly, was so shocking? "I no longer would have to travel to locations," he explained. "If I wanted to be in the snow in Colorado, it's text. If I wanted to write a scene on the moon, it's text, and this AI can generate it like nothing. If I wanted to have two people in the living room in the mountains, I don't have to build a set in the mountains, I don't have to put a set on my lot. I can sit in an office and do this with a computer."

AI Films as the Next Indie Films

Perry was clearly despondent about the future of his industry. "I am very, very concerned that in the near future, a lot of jobs are going to be lost," he told The Hollywood Reporter. "There's got to be some sort of regulations in order to protect us," he added separately. "If not, I just don't see how we survive."

But just think for a moment about his warning: "I can sit in an office and do this with a computer."

Perry's break into Hollywood and rise to moguldom was partly a result of discrete creative successes, particularly his Madea films, which targeted black audiences that mainstream Hollywood studios underserved, ignored, or sneered at.

But those movies also succeeded because of clever cost cutting. Perry's cinematic oeuvre is replete with movies shot quickly and for relatively small budgets. By making movies on the cheap, Perry had dramatically reduced the risk of filmmaking, especially for outsider filmmakers without an existing foothold in Hollywood. Perry's worry was essentially that AI would do something like what he'd done himself—but even more so.

The history of Hollywood is a history of grand visions and clashing egos, of storytelling triumphs and marketing coups, of penny-pinching sleeper hits and wildly overbudget failures.

But it is also a history of technological evolutions and business revolutions. From the dawn of sound to color to computer-generated special effects to digital photography and immersive sound effects, innovative filmmakers have always found ways to make advances in technology work for them. On the business side, shifts in distribution, marketing, and financing helped keep the risky business of making movies afloat.

The independent film boom that reshaped Hollywood in the 1990s began with the 1989 debut of director Steven Soderbergh's Sex, Lies, and Videotape at the Sundance Film Festival. Sundance had been a sleepy gathering that prioritized worthy-but-dull stories of regional Americana; even the most successful rarely played in more than a couple dozen theaters across the country. Sex, Lies, and Videotape played on hundreds of screens, including art-film-averse suburban multiplexes and malls. It grossed over $36 million worldwide on a $1.2 million budget, and turned Sundance specifically, and independent film generally, into a big-money bonanza.

Soderbergh, meanwhile, went on to be one of the most prolific, most influential filmmakers of his generation. Sometimes he directed conventional star-powered box-office smashes, such as Ocean's Eleven and its immediate sequels. But he also embraced technological change. In 2005, he released Bubble, a tiny-budget, almost experimental film released simultaneously in theaters and on digital streaming platforms, making him the first major filmmaker to do so. In 2019, he created High Flying Bird, a movie shot entirely on an iPhone and released first as a phone app with interactive elements. Throughout his career, Soderbergh has kept costs down by serving as his own cinematographer and editor.

Peter Biskind's 2004 book about the 1990s indie film renaissance, Down and Dirty Pictures: Miramax, Sundance, and the Rise of Independent Film, leads with a quote from one of the leading figures of that era's boom, Quentin Tarantino: "In the '80s, the studios could predict what worked and what didn't. And that's what the '80s were—one movie you'd already seen after another. Suddenly, that's not working anymore….When the audience is fed up with the standard stuff and crying out for something different is when exciting things happen in Hollywood."

Tarantino was talking about a different era. But something like that description applies to Hollywood after the pandemic, as superhero movies that should have been sure things flop and proliferating streaming services struggle to find subscribers. Generative AI, combined with the cheap, accessible, mass distribution of online video via platforms such as YouTube, could be the "something different" that drives the next era.

Entertainment Strategy Guy (ESG), the pseudonymous author of a newsletter about the business of Hollywood, suggests AI tools could birth a new role that he calls "the creator."

In this future, "people are not only writing their own scripts, maybe with assistance from AI, and directing and editing, but…also doing all the cinematography, special effects, and acting if the technology gets good enough," he says. It's a world in which "everyone has a chance to be their own Soderbergh" and "anyone can create their own masterpiece."

Hollywood, long the province of large and expensive creative teams, starts to look more like the world of book publishing, where single authors, perhaps working with an editor, are the norm. This is a future that advantages digital creators and independent producers, outsiders with few connections and little money. "For them, I think they see a lot of upside to it, because you can just go make your movie now," particularly if it's animated, ESG says in a phone interview.

But, he warns, there's a risk to this as well. The movie industry has already changed dramatically in recent years, thanks to advances in digital technology and online distribution, with streaming services such as Netflix and video hosting platforms such as YouTube weakening Hollywood's previous grip on video entertainment. His worry is that AI will supercharge this transformation. If, he says, "by the end of the decade Hollywood won't be making $200 million movies because they are all being made online—that's tens of thousands of job losses that could happen extremely quickly. And that disruption is extremely painful."

Although he sees optimistic scenarios as possible, he calls himself an "AI pessimist," in part because of AI's disruptive nature and the speed at which it might wipe out entire categories of work. He argues for attempting to slow down the pace of change to make it easier for today's film industry workers to adjust. But that simply may not be possible.

Every Person a Filmmaker

The integration of generative AI into filmmaking will be disruptive. It will result in large shifts in the nature of entertainment business employment. Job loss is difficult in the best of circumstances, and no one should make light of layoffs, especially when they severely damage an entire industry or sector. Yet we've seen large-scale job losses in a significant entertainment sector not too long ago—and sometimes those losses can pave the way for something better.

Unlike the young auteurs who reshaped Hollywood in the 1970s, Quentin Tarantino didn't graduate from film school. Instead, he clerked at a Manhattan Beach, California, video rental store. At the time, home video was a new technology; before VHS, cinephiles had to catch revival screenings at local repertory theaters or wait for movies to come on television. Tarantino was already a devoted moviegoer. But access to a vast VHS library helped Tarantino and others from his era gain a then-novel encyclopedic knowledge of and access to cinematic history.

Not only was his own cinematic education made possible partly by a then-new distribution technology, but his movies also found their audience in part via the home video boom, which, thanks to the popularity of VHS and then DVD, made more movies more accessible to more people.

Blockbuster, the biggest of the video rental chains, began with a single store in Dallas, Texas, in 1985. By the store's peak, according to Business Insider, the chain included more than 9,000 outlets in the United States alone, and employed 84,000 people globally.

Today, Blockbuster is defunct. Those jobs—some hourly service jobs, some managerial, some corporate and executive—are all gone. Aside from a handful of nostalgic holdouts, home video rental as we used to know it is lost. But streaming services and digital rental options make watching movies at home easier and more accessible than ever.

Something similar seems likely to happen to the rest of Hollywood. The most dire predictions of industry job cuts may not come true, but many of today's film workers are likely to be displaced by AI workflows that are some combination of faster, easier, more convenient, more flexible, and more accessible.

For some, that shift will be a cause for despair. In February, Colin Matthews, who has worked in various capacities on film sets, put up a droll post on X (formerly Twitter) portraying the ever-increasing expectations placed on those who want to be screenwriters. In 2004, he wrote, someone who wanted to be a screenwriter needed a "spec," or sample, script for an existing show. A decade later, that aspiring writer needed an original pilot. By 2024, the hopeful screenwriter needed not only a pilot, but a series bible—basically a creative and stylistic guide to the production of the show. By 2025, he joked, the same writer would need "a complete AI pre-viz"—a sort of animated test run—and by 2026, the punchline: "You need to make the show yourself."

This was presented as an absurd, comic vision of impossibly escalating demands on aspiring filmmakers. But there was an unintentionally hopeful reading, too: The promise of AI is that "make the show yourself" is, in fact, exactly what you'll be able to do.

AI tools won't just make big movies easier and cheaper to make. They'll give small filmmakers—perhaps even teenagers sitting in their bedrooms with their laptops—the tools to compete with better-financed professionals, leveling the playing field. The art and business of making movies will change and adapt, as it always has. And as always, opportunities will abound for those clever and imaginative enough to take them. An analog star like Tom Cruise will still be able to defy gravity and good sense with his stunts. But the digital Toms of the world will get a chance too.