Disney’s AI Gamble Is Riskier Than You Think

Disney’s AI Gamble Is Riskier Than You Think

For as long as there has been technology, there have been attempts to let ordinary people shape stories with it. The proto-interactive 1960s project The Sumerian Game allowed users to make choices in an ancient setting that a system could then react to and incorporate. During the 1990s, developers began working on Facade, in which…

For as long as there has been technology, there have been attempts to let ordinary people shape stories with it. The proto-interactive 1960s project The Sumerian Game allowed users to make choices in an ancient setting that a system could then react to and incorporate. During the 1990s, developers began working on Facade, in which a player could talk to characters in conflict and, with help from an early form of AI, alter the characters based on what was said. And in the 2010s, a company called Interlude allowed people to change dozens of “channels” on a video for Bob Dylan’s “Like a Rolling Stone,” shifting the story with it; a few years later, Black Mirror: Bandersnatch would expand the concept to feature form.

But something new brews now. Unlike past efforts, which required a large investment of money and professional know-how — the everyday “shaper” was always working within the boundaries that credentialed programmers set up for them — the current AI models dangle a more subversive idea. In this new system, anyone can initiate a take on the story, in any direction, thanks to models trained on millions of pieces of media. The programmers have left the building, and the guardrails have fallen from its scaffolding.

It is this scenario that Disney chief Bob Iger sketched out when he revealed Nov. 13 that the company would soon bring generative AI to Disney+ for subscribers to use, presumably in (on?) Disney+ movies and TV shows. “User-generated content,” as he quaintly put it, would land on the platform in the near future. Given how Sora 2 has been showing the ease with which known personalities and moments can be recast — just a rough prompt yields a passably coherent minute of video — Iger’s promise hardly seemed idle. The number of adults even over the age of 65 who have heard “a lot” about AI has doubled since 2023, according to a new Pew Research study. If Disney has its way, many of them, along with their younger co-subscribers, will be using it too.

The country’s biggest pure-play entertainment firm saying it will unleash AI tools on its nearly 200 million global streaming customers (whether using its own tech or someone else’s) marks a seismic moment. In 2017, Iger announced, years after Netflix began eating his lunch, that it would try to grab back some of the sandwiches. Thus Disney+ was born. With this AI announcement, the company signaled it won’t make the same mistake twice, becoming the first major entertainment firm to go all in on AI personalization. Any illusion from Disney’s ongoing lawsuit with Midjourney that it was skeptical of AI has been washed away; what it’s skeptical of is not controlling the AI itself.

We shouldn’t be misled by the length of these clips, either; Iger cited the play as “mostly shortform.” He was partly hedging on a tech not yet evolved (AI starts to fumble continuity after a minute or two). But mostly he was trying to cast the move in TikTokian terms — a rubric audiences understand and Wall Street craves. In truth, there is no reason these offerings won’t take a variety of formats. Professional shortform content pales financially compared with the longer stuff, and so as the technology gets better, the videos will almost certainly get longer.

On one hand, of course, this is the next step on an inevitable escalator to the level of hyper-personalized automated content. But on the other hand, a certain irony ripples though the move. Just six years ago, Disney touted massive spends on original content for its new streaming service, and so it can seem strange to hear Iger offer as his selling point the idea that you could now rip it up.

The consequences of this AI video moment go well beyond Disney. We are slowly becoming accustomed, cringey viral video by cringey viral video, to the idea that stories and personalities are not fixed entities, there to be interpreted as one likes but little else. Instead, AI companies condition us to think of them as far more malleable. However loathsomely some use the tech, to have Martin Luther King Jr. give a racist speech or place Stephen Hawking in an MMA ring, as two grotesque Sora spectacles had it recently, it still accomplishes something sneakily subconscious: instilling the belief that well-established narratives don’t have to remain static. We can switch them up at will. In fact, OpenAI makes it look fun.

Disney’s challenge is how to port this mindset to established properties. It seems cool to watch the Star Wars movies with faces grafted on to Darth and Luke or to drop a lightsaber battle into the Grand Canyon or your living room. But studios also face a challenge: These stories are so fixed in our consciousness that altering them can feel ick (as an online backlash to the announcement suggested) or at least uninteresting; better to use AI to create new concepts and stories. In which case, of course, we don’t need Disney at all.

The Hollywood C-suite recognizes this giant matzo ball at the center of the Great AI Promise and is here to reassure Wall Street that what audiences will want to use AI for is in fact to toy with “existing IP,” or what ordinary people call our cultural memory. “AI is going to give us the ability … to provide users of Disney+ with a much more engaged experience, including the ability for them to create user-generated content,” Iger told investors on a recent conference call. He may be right. But the full appetite has yet to be proved.

For all the drama attending the AI announcement, it remains deeply unclear how people will use it. Will we deploy the tech to create new endings, alternate scenes, fresh sequels or even entirely new story content that eschews the professionals? Or are we heading to a world that’s a lot like this one, of quick memes and inside jokes, of personal adornment and cameo-ized appearances, only now steroidal thanks to AI? Nobody knows, and, like the tech companies’ own relationship with their models’ inner workings, nobody seems to want to know. Better to just do it and see whathappens.

Confronted with all the ways it will mess with and potentially undermine artistic originality, AI cautionaries like to cite Jeff Goldblum’s classicIan Malcolm line, “Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.” When it comes to video generation, at least, Malcolm’s warning doesn’t seem much heeded these days. And pretty soon, at tonight’s family-room screening of Jurassic Park, it may be something he doesn’t say at all.