At the risk of belaboring the obvious, generative AI is now everywhere in the media and rights-based industries. It’s writing news articles and fan-fic e-books, it’s making music, it’s creating artwork. But no creative industry will be transformed by AI quite as much as movie and television production. The reason has as much to do with economics as technology.
Warner Bros.’ “Dune: Part Two” opened to a whopped $81.5 million domestically over the weekend, and $97 million internationally. It brought a welcome boost to theaters, which had seen the number of butts in seats come crashing down from the summer’s “Barbenheimer” high. And it showed that big-budget, effects-driven spectacles can still deliver for a studio, especially if they’re spectacular enough to justify release on large-format screens, like IMAX, which carry a premium ticket price and accounted for 48% of “Dune’s” domestic tally.
But blockbuster hits like “Dune,” “Barbie” and “Oppenheimer” are few and far-between these days. Prior to this weekend, the domestic box office was running 20% behind the first two months of 2023, and even Dune 2’s impressive haul only managed to cut that deficit to 13%. The highest-grossing movie in the fourth quarter, “Taylor Swift | The Eras Tour,” wasn’t even a studio film. It was financed by Swift herself and released a distributed by theater-operator AMC Entertainment.
In any case, a handful of blockbusters sprinkled throughout the year isn’t going to fix all that ails Hollywood today. That’s because the heart of the problem lies on the cost side of the ledger, not the income side.
The studios spent $gazillions over the last four years building out their own proprietary direct-to-consumer streaming platforms, partly in response to the Covid pandemic but mostly in pursuit of the kind of tech-company like share price multiples they saw Netflix earning. Only later did they realize the full cost of feeding the beasts they had created, or begin to appreciate the anti-synergistic tension between their theatrical distribution and streaming businesses.
Even as subscriber rolls increased, which boosted streaming revenue, NBCUniversal managed to lose $2.75 billion on Peacock in 2023. Paramount Global posted a $238 million streaming loss in Q4 alone and has put itself on the block, and Disney, while managing to somewhat staunch the streaming bleeding, still lost $387 million on Disney+ and ESPN+ in the quarter.
TV show budgets today can range as high as $25 million per episode. Movie’s like “Dune: Part Two” can easily cost $200 million or more to produce and market. In the face of fundamental changes in audience behavior wrought by Covid and by the studios’ own conflicted distribution strategies, those cost are no longer sustainable, especially for companies with debt-laden balance sheets left over from a decade of frenzied (and often ill-conceived) mergers and acquisitions.
That’s where AI comes in. Far more telling about the future direction of Hollywood than the successes of “Barbenheimer” and “Dune” is the reaction by actor/producer/writer and director Tyler Perry to being introduced to OpenAI’s Sora text-to-video generator. After seeing a demo of the generative AI tool Perry immediately halted work on a planned $800 million expansion of his studio facility in Georgia. “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. It’s shocking to me,” he told The Hollywood Reporter.
That’s not likely to be the last aborted investment in traditional production infrastructure. AI is already widely (if quietly) in use in Hollywood. It has been used in script evaluation for several years, and is finding its way into VFX suites. As the technology continues to improve it will likely prove a significant, possibly essential, cost-savings tool as the cost of traditional means of production grows increasingly unsustainable.
The writers and actors who shut down movie and TV production last summer were not wrong to fear for their livelihoods at the hands of genAI. The development of realistic, high-quality video from a few text prompts is only going to accelerate AI’s adoption. Director James Hawes, who worked on “Slow Horses” and “One Life,” recently described fully AI-generated TV shows as “inevitable” within 3-5 years.
To be sure, AI is in use and will continue to be in use in other creative industries, including music and journalism, and may ultimately displace some human creators. But none of those industries have anything like the cost structure to propel adoption of movie and TV production. I’m performing a kind of journalism as I write this, but it is not costing (or, sadly, yielding) me millions of dollars.
Moreover, there are aspects of those crafts that even today’s most powerful AI systems cannot reproduce. A music generator can write a song and an audio generator can produce it. But for now, at least, an AI cannot perform live or induce a fan’s emotional response to an artist. A Large Language Model can generate a news article from data it is fed, but it can’t yet generate that data itself. It can’t meet a background source in an underground garage at night, or coax a leak out of a reluctant official.
In contrast, there are many synthetic elements we accept in movies and TV shows without blinking an eye, from CGI and synthesized sound effects to virtual sets and backgrounds. A willing suspension of disbelief is an essential part of the moviegoing experience, in fact.
That difference also partly explains why the studios have stood apart from their usual allies in other creative sectors on the question of copyrightability of AI-generated content. While the music and publishing industries have been adamant on denying copyright protection to synthetic works, the studios have urged policymakers not to throw out the copyright baby with the AI bathwater.
“MPA is troubled that the Office is moving toward an inflexible rule that will deny registration if human users are not able to predict and control the particular outputs that follow from prompts provided to the AI system, despite extensive human involvement in the creative process,” the Motion Picture Association said in comments submitted to the Copyright Office in its inquiry into copyright and AI. “Even if such an approach is appropriate for some uses of ‘generative AI’ systems like Midjourney, the approach should not apply to MPA’s members’ use of AI as a production and post-production tool.”
Here come the movie bots. Anything an AI can put on screen will be on screen.