Why Is an Amazon-Backed AI Startup Producing Fan Fiction About Orson Welles?

On Friday, a startup called Fable introduced an ambitious and, at times, perplexing project aimed at reimagining the 43 missing minutes from Orson Welles’ legendary film “The Magnificent Ambersons.”

What drives a startup that describes itself as the “Netflix of AI,” especially one recently backed by Amazon’s Alexa Fund, to take on a film that’s over 80 years old?

The answer lies in Fable’s innovative platform, which allows users to generate their own animations using AI prompts. While the company initially focuses on original creations, it aspires to extend its reach to include classic Hollywood films. In fact, its technology has already produced unauthorized episodes of “South Park.”

Fable is launching a unique AI model developed to create intricate and extended narratives. Over the next two years, filmmaker Brian Rose — dedicated to reconstructing Welles’ original vision for five years — plans to use this model to recreate the missing scenes from “The Magnificent Ambersons.”

It is essential to note that Fable has not obtained the film’s rights, suggesting that this initiative may function more as a showcase for the technology rather than a work intended for public release.

Why focus on “Ambersons”? If you’re not familiar with Welles, this choice might appear odd for a digital revival. Among film lovers, “Ambersons” is often overshadowed by its more renowned counterpart — while “Citizen Kane” is frequently celebrated as the greatest film ever made, Welles’ second feature is often regarded as a lost gem, drastically altered by studio executives into an implausibly upbeat conclusion.

The film’s legacy — delving into themes of loss and unfulfilled potential — likely piqued the interest of both Fable and Rose. However, it’s vital to emphasize that our contemporary fascination with “The Magnificent Ambersons” largely stems from Welles himself, particularly given how its incomplete release adversely impacted his Hollywood career while still showcasing his directorial prowess despite its imperfections.

It is particularly significant that Fable seemingly did not consult with Welles’ estate. David Reeder, who manages the estate for Welles’ daughter Beatrice, described the project in Variety as an “attempt to gain publicity using Welles’ creative legacy,” contending that it would ultimately be “a purely mechanical endeavor devoid of the unique innovative insights a creative mind like Welles would have brought.”

Despite Reeder’s criticism, he appears more troubled by the lack of communication from Fable than by the project itself. He noted that “the estate has embraced AI technology to develop a voice model intended for use in voiceover work with brands.”

Personally, I’m less enthusiastic about this concept. Even if Welles’ heirs had been consulted and compensated, my interest in this new “Ambersons” would remain minimal, similar to my reluctance to hear a digital mimicry of Welles’s iconic voice promoted in marketing campaigns for new products.

Welles enthusiasts are aware that this is not the first instance of his films being edited or completed posthumously. However, previous attempts at least utilized footage originally shot by Welles. In contrast, Fable proposes an approach that blends AI with traditional filmmaking — reports indicate that specific scenes will be reshot with contemporary actors, whose faces will then be digitally modified to resemble the original cast.

The absurdity of announcing such a project without film rights or Welles’ daughter’s consent is evident, yet Rose appears genuinely committed to honoring Welles’ vision. For instance, when expressing his desire to recreate the film, Rose conveyed his sorrow over losing “a four-minute-long, unbroken moving camera shot, the absence of which is a tragedy,” with only 50 seconds of that shot preserved in the edited version.

I share his sense of loss — but I also believe this is a dilemma that AI cannot resolve.

Regardless of how expertly Fable and Rose reconstruct their rendition of that tracking shot, it will ultimately be their shot, featuring digitally rendered representations of Joseph Cotten and Agnes Moorehead, and not the actual actors. The finished product will not capture Welles’ true vision for “The Magnificent Ambersons” that RKO altered nearly 80 years ago. Without a miraculous rediscovery of lost footage, that version remains irretrievably lost.