THE ECONOMIST: Hollywood is entering its AI era as Netflix, Amazon and even human actors change their tune

Artificial intelligence is a touchy topic in Tinseltown. Rancour over the prospect of so-called generative AI being used to produce scripts and create digital replicas of actors was among the grievances that led to strikes in 2023. Hollywood hotshots have since continued to rail against the technology.

Kane Parsons, director of the hit film Backrooms, recently said that, if he could, he would “snap my fingers” and make AI “disappear”.

Christopher Nolan, creator of a new blockbuster adaptation of The Odyssey, describes it as a Trojan Horse, arguing that the motives of those championing it should be viewed with scepticism.

In May Amazon MGM, a studio, approved an AI-supported animated series called Punky Duck. Two days later the director abandoned the project after receiving death threats.

Yet behind the scenes AI is quietly playing a growing role in every aspect of film-making, from script development and production planning to shot generation and editing.

In March Netflix acquired InterPositive, an AI production company led by Ben Affleck, for almost $US600m. Last month the streaming giant said that it had used AI in the creation of about 300 titles this year alone.

In June Google DeepMind, an AI lab, invested about $US75m in A24, the studio behind Backrooms, as part of a research partnership to develop AI tools.

So far there is little evidence that AI is about to replace humans on the red carpet — though the much-publicised arrival last year of Tilly Norwood, an AI-generated actor, did cause a fleeting moment of panic.

“It takes great artists to make something great, and AI is not changing that,” argued Ted Sarandos, Netflix’s co-chief, in July. Instead, the technology “will give creators better tools to bring their visions to life.” Consumers will be all the better for it.

Peter Woodbridge, author of a book on AI and entertainment called Dream Machine, says that many studios are currently taking a “don’t ask, don’t tell” approach to using AI.

But some filmmakers are less coy. In his latest film, Young Washington, director Jon Erwin used the technology to superimpose a raging torrent on a 50-foot trench, sparing his actors a potentially dangerous stunt.

Mr Erwin says he finds the technology “creatively addictive”, and stays up late into the night experimenting with it.

Netflix’s first use of generative AI in film-making involved simulating a collapsing building in last year’s Argentine sci-fi series The Eternaut. Its recent documentary series, The American Experiment, included 17 minutes of AI-enhanced footage including historical battle sequences, which Mr Sarandos has said were produced “twice as fast and at half the cost” of other options.

Eline van Der Velden, creator of Tilly Norwood, guided The Economist through creating a ten-second clip of British police arresting the AI actor. The scene, which took four minutes to generate, would have taken half a day to shoot on location, says Ms Van Der Velden — though she adds that more takes with the software would be needed to generate a professional clip.

Mr Erwin says he hopes that AI will upend the “unsustainable economics” of Hollywood, where huge costs lead to risk-averse choices such as the creation of endless sequels.

Others fear a deluge of AI-generated slop, and point to the crassly titillating “micro-dramas” coming out of China, which increasingly use the technology, as a cautionary tale.

But Evan Shapiro, a TV producer, is sanguine about the risk such “bullshit” presents to Hollywood, arguing that viewers will continue to appreciate high-quality content.

Indeed, harnessing AI may help studios better compete with micro-dramas, as well as other video platforms such as YouTube and TikTok that have been winning the attention of audiences.

Mr Shapiro calculates that viewers aged 13-54 in America and seven other major markets now spend more time perusing YouTube than watching the content produced by Disney, Fox, Paramount, NBCUniversal and Warner Bros Discovery combined.

That is spurring the film industry to use AI to repackage its content into shorter clips optimised for consumption on a smartphone.

Social Department, an LA-based startup, is using the technology to help studios give old series new life by slicing them into scenes that are distributed through social media. Jonathan Verk, its founder, notes that AI is particularly useful in the fiddly task of turning video shot in a horizontal format into a vertical one suitable for mobile.

Tinseltown’s stars are also beginning to change their tune on AI. Alexandra Shannon of Creative Artists Agency, which represents many A-listers, says that at the time of the strikes in 2023, a number of her clients were concerned with how they could protect their likenesses and voices from exploitation by model-makers. Now they are excited by the opportunities, provided they come with what she calls the “three Cs”: consent, controls and compensation.

There’s nothing you can do about it. It’s the future.

Still, copyright remains a sensitive point. In February a video went viral showing Tom Cruise and Brad Pitt fighting on a rooftop in Los Angeles.

The strikingly realistic clip had been produced using Seedance 2.0, a model from Bytedance, the Chinese company behind TikTok. Disney, Paramount and the Motion Picture Association, an industry body, were among those who sent Bytedance letters demanding that it cease its alleged copyright infringement.

Bytedance, which insists that it respects intellectual property, responded by adding new safeguards to its video tool. In December Disney signed a deal with OpenAI, maker of ChatGPT, to license characters such as Mickey Mouse for use in its Sora video model. But the deal collapsed in March when OpenAI shut the model down.

Silicon boulevard

Big tech’s encroachment on Hollywood may help to encourage the dissemination of AI through the film industry. Amazon, whose name crops up surprisingly often, provides the clearest illustration.

Along with its ownership of MGM, it exerts influence through its Prime Video streaming platform and Amazon Web Services (AWS), which provides cloud infrastructure to many of the big studios. Samira Panah Bakhtiar, an AWS executive, calls it the “invisible backbone” of entertainment.

For example, her business works with Fox Sports to convert broadcasts into vertical form at low latency for smartphones. The ill-fated Punky Duck series was supported by the GenAI Creators’ Fund, a joint initiative by AWS and MGM to encourage the adoption of AI tools in film-making.

AWS’s efforts to promote AI in Hollywood may stem in part from the high computational-intensity of video tools, which could help it to generate a return on its colossal investment in AI data centres. Google, whose Veo family of audio-visual models are popular in Hollywood, may be investing in them for similar reasons.

Even so, there will be plenty of hold-outs. When Mr Erwin got his start in the industry in the 1990s, he used a $US3000 digital camera rather than a 35mm one costing $US300,000. At the time, established filmmakers saw digital technology as a threat to the prestige of their craft. Now, for all but purists such as Mr Nolan, it is commonplace.

Eventually AI, too, may become an uncontroversial aspect of filmmaking. In a recent interview George Lucas likened opponents of generative AI in his industry to people who once railed against cars because they were more dangerous than horses and carts.

“There’s nothing you can do about it,” he said. “It’s the future.”