A screenwriter friend once told me about a young assistant who was handed a script by his boss and told to drive it across town, to Bel Air, to the stately home of one of the world’s most celebrated directors. He walked up the long driveway, past the manicured gardens, before handing off the script. As the director inspected it, the assistant said nervously, “I have to say, your house is just incredible.” Without skipping a beat, the director shot back, “Yeah, well, no one who lives in it is happy” and slammed the door. When I asked why the director was so miserable, my friend replied: “Probably because he works in Hollywood.”

One reason the industry is so much more taxing than other creative fields is because it’s so expensive to make anything. According to producers and studios I’ve spoken to, TV show budgets now range between $6 million and $25 million an episode, not including marketing costs. Most mainstream movies now cost between $100 million and $250 million to make. Years ago, you could make a blockbuster for a fraction of that. The first Top Gun (1986) cost $15 million to make. The 2022 sequel cost $170 million.

But this is all about to change because of AI. This past month, OpenAI announced Sora, which can take text and turn it into astonishingly realistic video, in the same way the company’s other products, like ChatGPT, can with text-to-text, or Dall-E can with text-to-images. Days after the announcement of Sora, the media mogul Tyler Perry said he was stopping an $800 million expansion of his Atlanta film and TV studios. “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.” (Perry acknowledged in the same interview that he’d used AI in two upcoming films.)

There are also platforms like Pika, Runway, and VideoPoet, made by Google, which offer competing text-to-video AI software that can create short clips in whatever style you want. These technologies can make video from text or images, taking a still image and animating it in a way that makes it look like it’s a scene out of a $170 million production. Go look at the demo reel for Wonder Studio, an AI special effects company that uses drag-and-drop to change an actor into, say, a robot or an alien, to see just how quickly these advancements are happening.

It’s not just visual effects that can be done with drag-and-drop algorithms. It’s everything. Text-only LLMs, like Squibler, Jasper, and ChatGPT, can already write mediocre scripts. The same is true for start-ups that enable you to create the score for a film using a full philharmonic of brass, woodwinds, percussion, and strings. Then there’s the AI editing platforms that can string it all together. All of this hints at a not-too-distant future where a film or TV series could be made by a single person, though regrettably that single person would not be Mike White.

While most people who work in Hollywood—no doubt including the 90 percent who are already struggling to make a living—don’t look on these developments as desirable, there’s a swath of people who see them as an inevitable addition to storytelling, at least one day. “Creators in all realms are going to have to look at this as a profound opportunity. They will be able to collaborate with a force that can sift through every grain of cultural capital on the planet and traverse the entire kingdom of history in less time than it takes to pour a cup of coffee,” says Allan Loeb, a screenwriter who is currently writing a novel about creative AI in Hollywood. “The apprentice’s 10,000 hours will become eight seconds.”

Putting aside the hotly disputed copyright issues, Hollywood is about to experience a disruption similar to what happened more than two decades ago in the music industry, where a person used to need access to an exorbitantly expensive recording studio to make a single song (not to mention agents, managers, and distribution deals), until the MP3 and inexpensive software let artists like Justin Bieber (who was discovered on YouTube) and Billie Eilish (who was discovered on SoundCloud) subvert the usual channels of music stardom. Today between 25,000 and 100,000 new songs are uploaded every 24 hours to Spotify, mostly made, presumably, by people in their bedrooms. Your guess is as good as mine as to how many of these songs are actually any good or how many people actually listen to them. But imagine when anyone can create their own Oppenheimer-length film that could pass for the product of a big studio.

“By giving everyone access to AI tools that will allow individuals to make films, music, animation, and more, we will open up these mediums to whole populations of people who would otherwise never have the possibility of telling their story,” says Mika Johnson, a filmmaker and documentarian who also works in AI. The problem is that any idealized Cambrian explosion of access is almost certain to collapse on itself. “Every artist on the planet is having a Wile E. Coyote moment,” Loeb says. “My only advice to them is that they may not want to look down.”

Hollywood is a town in a perpetual existential crisis, going back to the transition from silent films to talkies in the ’20s, to the end of the studio system in the ’40s, to the rise of television in the ’50s. This past decade has pulled and pushed and slapped Hollywood in every which way possible. The streamers broke those lush business models, and then COVID-19 ground the industry to a halt. And right when it was about to get greased up and ready to churn out content again, the WGA and SAG went on strike.

There was a long list of demands from the unions, from residuals to data, but at the heart of the fight last year, the unions wanted to ensure that their members wouldn’t be replaced by AI. The studios agreed not to use AI in lieu of writers and mainstream actors. But three years from now, when the next negotiations begin, AI video and AI writing technologies may be a thousand times more advanced and ubiquitous, and that battle will be even harder to win.

Many people I’ve spoken to who work in AI say, like Johnson, that their goal is to democratize creativity, making filmmaking and storytelling available to anyone. The aforementioned demo released by Wonder Studio, which boasts an advisory board of industry heavyweights like Steven Spielberg and Joe Russo, emphasizes how its tools are meant to empower artists rather than replace them. And I imagine most writers, directors, producers, editors, and cinematographers don’t believe that a machine can do what they do. But the shift is already happening.

Today, language dubbing is possible and in some instances is already being done by AI voices and deepfake technology that can not only make Tom Cruise speak fluently in Spanish or Russian or Leonese, but can also make his mouth move in sync. (Likewise, Kendall Roy manipulated video of his dead father on Succession; in real life, the internet recently witnessed an insidious version in the form of deepfake explicit images of Taylor Swift.) The rest of the industry will transition gradually. Some short clips will be created for films and TV shows that were made by AIs, perhaps B-roll or pickup shots. Then extras will be replaced by AI actors—in many cases, extras have already been replaced by CGI. Then temporary music made by AI composers will become the actual compositions for films and TV. Writers rooms, bastions of collaboration in all its inefficient human glory, will be culled, one at a time, replaced by AI writers.

Andy Weir, author of The Martian, which was made into a feature film nominated for seven Oscars, is a true believer. “AI will be able to write much better stories than humans can. Mainly because it will learn what you do and don’t like, and it’ll write a story tailored to be something you will enjoy,” he says. The scenario goes like this: For a while, humans will still be involved. Then the AIs will start to make content just for you—and only you—based on how you feel that day. The cameras in your phone and laptop and TV will be able to see your heart rate and emotions (AI can already do this), calculating all those neurons as they click and clack in real time, and then change the content as you’re watching to make it funnier or more emotional, or remove characters you don’t like. “I also think,” Weir says, “that—before my life is over—my profession will effectively disappear.”

Tim Cook, CEO of Apple and a big part of the reason we all carry camera-wielding phones in the first place, begs to differ. “You think about the great storytellers,” he tells me. “Would you rather watch something from them, or do you want to watch something that is AI-generated? I don’t think it’s going to change in 5 years or 10 years.” Most of Hollywood will be glad to hear it.