How a Netflix Doc Used AI Winston Churchill — For the Better (Really!)
I talk to Imagine's Justin Wilkes and Sara Bernstein, producers of 'Churchill at War', about how their series fights the narrative about AI
Erik Barmack writes every other Tuesday for paid subscribers. He recently considered how AI-powered humanoid robots could replace film crews, how AI storyboarding tools transform pitching scripts and whether Hollywood should take AI money.
“The price of greatness is responsibility.” So said the booming Winston Churchill during a time when democracy was under duress and the world was precariously immersed in a global war.
But have we actually heard him say it? I can’t find an online recording of a line that I have imagined hearing for years — only transcripts of this famous speech, given at Harvard in 1943.
In the new four-part docuseries Churchill at War (debuting on Netflix on Dec. 4) — a series that required me to slip on my dad jeans and put down my latest Lincoln bio in anticipation — the pugnacious statesman delivers the line, and it’s a triumph of, yes, the magic of the latest AI technology and not an archaeological discovery.
The documentary producers, Imagine Entertainment, brought Churchill to life by taking hundreds of speeches that have only been available on paper to historians and his legion of fans and giving them voice as if they’d been recorded. They also colorized photos and videos of the Prime Minister, which also creates a kind of vibrancy to understanding Churchill during the time when the Prime Minister protected his country from the greatest threat it ever faced. “What we started to dabble with,” Justin Wilkes, Imagine’s president, tells me, “could we essentially have Churchill narrate his own story? Could we actually put Churchill in the first person to take us through as the audience and get that insight as to what he was thinking every step of the way?”
Yet the space of bringing a voice back from the dead is not without its ethical challenges. As Churchill also wrote, “A lie gets halfway around the world before the truth has a chance to get its pants on.” About a month ago, a Polish radio station, Off Radio Krakow, broadcast a new interview with Nobel Prize winner Wislawa Szymborska. She was probably unaware of it . . . because she died in 2012.
The poet was created using AI and her interviewer was an AI-generated Gen Z pop-culture reporter named “Emilia” and ChatGPT generated the conversation. This was bold. But it was also wrong and deepfake-y. In an attempt to be a “pioneer” and revive the shrinking business that is radio, the station fully invented a piece of “history.”
Documentary filmmaking and unscripted content are at the forefront of how AI is being used to tell stories. In our world of deepfakes and cheap fakes (AV manipulations to recontextualize footage), adopting AI is not without tradeoffs or the risk of controversy. But what makes Churchill at War worthy of a deeper look is that unlike several pieces of content we’ve covered that are intended either to simulate reality or create shortcuts, this series is a mainstream piece of entertainment streaming on the largest service in the world. It uses AI not as a gimmick or cost-cutting measure, but rather as a tool to improve the quality of the story being told. Churchill at War is, dare I say it, more authentic for its use of artificial intelligence.
To find out not only how Imagine did it but how they wrestled with the ethical issues presented by the use of AI in documentary, I chatted with both Wilkes and Sara Bernstein, head of Imagine’s documentary division.
In this article, you’ll learn:
The specific tools Imagine used to bring Winston Churchill back to life
How the producer created jobs in its use of AI
The ethical considerations Imagine took into account — even before the Archival Producers Alliance (APA) issued its guidelines
How factual content creators have gotten into ethical hot water using AI
The tension between disclosure and creating immersive entertainment
The dangers presented by one voice AI tool in particular
Where the APA guidelines already fall short with the advancing state of AI tech
How you can experiment with these technologies yourself today, even in everyday life