Back in February, Hollywood received a message from the future: “I hate to say it, but it’s likely over for us.”

This came from Wolverine & Deadpool screenwriter Rhett Reese in response to a clip posted on Twitter by the Irish filmmaker Ruairi Robinson. The AI-generated video, as you’ve almost certainly seen, depicted a fight between Brad Pitt and Tom Cruise atop what appeared to be a Roman ruin set somewhat perplexingly against the New York City skyline.

The clip seemed to represent a huge leap forward in AI-generated video, with high production values and realistic facial expression and movement. It looked like a movie. An expensive one. While some commenters were critical, the quality of the video represented a quantum leap forward from the famously crude early AI videos depicting Will Smith eating spaghetti.

Anyone who extrapolated that progress into the future could see that it wouldn’t be long — a year, months, weeks — before anyone could generate movies on their home computer that once cost $200M or more at no cost. For screenwriters like Mr. Reese and myself, and for everyone else who earns their living in our business, the prognosis seemed clear: It’s over for us. 

Or is it? Because the question, I think, isn’t what AI can create. The question is whether anyone will watch it. 

One harbinger of the paradigm shift we’re going through — not just in the entertainment industry but across our entire culture — took place May 11, 1997, when chess grandmaster Garry Kasparov lost to IBM’s Deep Blue. At that point, Kasparov was the greatest chess player in human history, with an ELO rating of 2800, but after that loss, computers left human beings behind and never looked back. To any observer on that day, it would seem that chess, or chess played by human beings, was over. The future would be computers playing other computers.

But that’s not what happened. Competitive chess — human chess — is thriving. Magnus Carlsen is a global celebrity. Chess has made him wealthier than any player before him, and YouTubers like Hikaru Nakamura earn as much as Carlsen. Chess is an industry in a way that it never was in 1997. Meanwhile, there is a much smaller community of people who design AI chess engines and pit them against one another in what I like to imagine as Fight Club-style contests held in basements. The AI chess bots are now vastly better than any human player who has ever lived, but — and this is important — nobody cares about them.

Human beings are social animals. We care about what other human beings do. We don’t really care that much about what machines can do. We take it as a given that they can do things we can’t — after all, that’s why we made them. But they don’t interest us the way other humans do.

Imagine you’re a tennis fan and that in the future Netflix’s algorithm is able to generate tennis matches custom-designed to your taste. You can watch a new, spectacular, nail-biting match between AI Carlos Alcaraz and AI Novak Djokovic every day. Would you watch that channel? Of course not. You don’t care about AI Novak Djokovic. To be a sports fan is not merely to be invested in the spectacle of sports but in the human drama of it: the contest between a young player on his way up against an older one trying for one last title. We feel we know these people, and what’s crucial to the drama of that parasocial relationship is that they are real human beings, testing real human limits. 

We care about the real Novak Djokovic because he’s a 38-year-old man seemingly doing the impossible — just as we care about the real Tom Cruise performing his own stunts because he’s a 63-year-old man doing the impossible.

Now imagine you’re a movie fan. Netflix’s algorithm detects that you loved Sentimental Value, so it creates a channel that auto-generates a new Joachim Trier movie for you every day. Would you watch it? Probably not. Because you’re interested in what Joachim Trier has to say and how he’s choosing to say it. You have a parasocial relationship with him and his actors, even if you’re walking into the theater for the first time and you’ve never seen one of his movies before. He’s a human being trying to communicate something to you, and that communication is part of a broader cultural conversation that takes place over tables at restaurants and parties and on Twitter. And if human beings aren’t involved, we don’t care. 

To be sure, not all entertainment is auteur cinema, and not everybody cares whether what they’re watching was human-made. Would a 3-year-old know or care whether Teletubbies was made by AI? Probably not, and this has implications for children’s programming. And future generations of kids raised on AI programming — “AI natives” — will likely be less discriminating about whether their entertainment is AI- or human-made. There’s a whole stratum of “commodity TV” where the consumer just wants something on in the background, and we should assume that there will be a sort of cultural creep as AI slop becomes more pervasive and starts moving up the entertainment food chain. 

Reese worries that very soon, “a person will be able to sit at a computer and create a movie indistinguishable from what Hollywood releases.” But this medium has existed as long as cinema itself — it’s called animation. We may be on the verge of an era of photorealistic animation, but that won’t necessarily crowd out live-action filmmaking any more than Pixar did. Tellingly, animation has never tried to perfectly replicate what live-action can do: instead, it does what live-action can’t. And in all great animation, from Chuck Jones to Hayao Miyazaki to Wall-E, the human touch is palpable in every frame, because human audiences demand it.

The AI filmmaker of the future will have to create something enormously original and distinctive to rise above the ocean of zero-cost AI slop that will soon be out there. If anything, the bar will be higher for AI filmmakers than live-action filmmakers in terms of voice and originality. The “AI Chris Nolan” that Reese predicts will have to be very, very distinctive in order to avoid being instantly duplicated by legions of roving AI agents cloning whatever movie just dropped.

The future of AI video may not be in creating photorealistic “fake” Hollywood movies with fake stars but movies that Hollywood can’t make, or that it has heretofore never even dreamed of making. The true frontier of the medium may be the point where the merger of “real” live-action performance and AI make films that were once uneconomic suddenly possible.

The real significance of the Brad Pitt vs. Tom Cruise video is that we are fast approaching the point where the cost of producing empty spectacle is zero. Visual imagery that once cost $250,000 per shot and thousands of man-hours to produce in CGI will be as plentiful as air, or water. We will soon be swimming in it, and its cultural and economic value will decline accordingly. But the supply of human drama, written by real writers and performed by real actors, will remain as scarce and valuable as ever.  

David Scarpa has been a leading voice in film and television for more than 20 years. Represented by Verve, he most recently wrote Gladiator II and Napoleon, both for Ridley Scott.