Writers and actors aren’t the only people in Hollywood grappling with the impact generative artificial intelligence will have on the entertainment industry. documentarians are also concerned about AI and what it means for the ethical standards and practices of nonfiction filmmaking.
Many have used AI to transcribe interviews in the past few years and in recent months generative-AI models including ChatGPT and Midjourney have helped docu assistant editors create spreadsheets and visual placeholders as well as extract and catalogue metadata. But recent advancements in AI, such as the ability to generate fake photographs and only needing three seconds of someone’s voice to create synthesized audio of that person saying anything, have filmmakers like Dawn Porter (“The ‘Lady Bird Diaries”) worried.
“We are supposed to be the truth, and it might be the truth as we see it, but we are also supposed to be transparent,” says Porter. “I’m very nervous that people are not going to be transparent about what techniques they are using and why.”
In 2021 Morgan Neville came under fire for not disclosing that he used artificial intelligence to create three quotes with Anthony Bourdain’s voice in “Roadrunner: A Film about Anthony Bourdain.” Neville’s filmmaking morals were called into question, as were the ethical boundaries of the entire docu community. If Neville could slip in a deepfake, what else could documentarians do that wasn’t truthful?
Plenty.
In addition to deepfakes, and AI-created photography and music, there’s also the issue of AI-generated “Frankenbiting,” which makes it easier than ever before to edit clips and/or dialogue out of sequence.
“I’m old enough to remember when we went from film to digital,” says veteran doc editor Bob Eisenhardt (“Free Solo,” “The Rescue ”). “That tool was amazing. You could do so much more, so much faster, and therefore you could consider so many other possibilities. So, I think when it comes to AI, it can be made to really be an essential tool. That is terribly exciting and terribly scary.”
Andrew Rossi incorporated AI into his 2022 docuseries “The Andy Warhol Diaries” to recreate the voice of the Pop Star icon.
“Several factors led me to feel that using an artificially-generated voice would be not only acceptable but reinforce my creative aims for the series,” says Rossi. “Most importantly, Warhol’s estate agreed, and I was very upfront with the viewer, flagging the AI voice at the top of the show.”
Transparency, according to Rossi, is key when it comes to using AI in the documentary form.
“There is an ethical calculus in using artificial intelligence, and the best way to proceed is to be really intentional, solicit advice and most of all, be totally upfront with the viewer and your subject,” says Rossi. “But I believe there are really exciting creative possibilities with AI, and it is not a tool that should be off limits in documentaries as a genre per se.”
Gary Hustwit agrees. The “Helvetica” director is getting ready to unveil “Nothing Can Ever Be the Same,” a 168-hour-long documentary video installation which he co-directed with British artist Brendan Dawes. The film, created solely by generative-AI, serves as a companion piece to Hustwit’s upcoming documentary, “Eno,” about musician Brian Eno. Hustwit and Dawes used their own IP — interviews, voiceover, B-roll, stills, music — and put parameters in place so that generative software, which the duo developed, could create a film that never repeats itself.
“Think about all the different folders and all the media that we have on our hard drives,” says Hustwit. “Whether it’s video clips, audio music, whatever, the (AI) system we created basically pulls from those folders and matches things up. According to rules and parameters that we set up, it chains together, scenes and music. It can create a film like the generative doc we made “Nothing Can Ever Be the Same.” We have no idea what the film will be.”
“Nothing Can Ever Be the Same” will premiere in October in Venice at the La Biennale di Venezia.
Bryn Mooser, founder of XTR, the production company behind doc favorites including “The Territory” and “I Didn’t See You There,” says that the company is only using AI as a practical ingestion tool.
“I think that you’re going to see AI tools coming into the doc side of post-production very soon,” Mooser says. “It’s certainly going to be the most disruptive thing to our industry since the computer probably. But the computer certainly was additive in the end to filmmaking overall.”
Nikita Liamzine concurs. As a post engineer at Imagine documentaries, Liamzine regularly uses generative-AI when editing a docu. He is confident that AI solutions in the edit process of a doc can drastically reduce menial job duties and assist human productivity.
“Ultimately the objective of an assistant editor is to become an editor,” says Liamzine. “Now that can happen more quickly because they can leverage those menial tasks by asking AI to help handle them. That will ultimately free them up to focus on more creative aspects of the editing process.”
Eisenhardt hasn’t used AI for any menial editing tasks, but he is currently working on a project that incorporates artificial intelligence. He wouldn’t name the film or how AI is being used within it, but Eisenhardt did admit that “AI can be a really exciting tool. It doesn’t interfere with the editing. It enhances the story.”
Mooser is hopeful that AI will continue to boost doc storytelling.
“Figuring out how we can use AI as a tool to augment what we’re doing and not to replace is ultimately going to be the big challenge that we all face,” he says.
But job replacement has become a growing concern within the doc field. Will there be a need for assistant editors and composers with AI in place?
“We are under such budget pressure because nothing is selling quickly,” says Porter. “Networks and streamers come to us and say, ‘Can you make it for X dollars less?’ And the way to make it for X dollars less is to cut out the people. The most expensive part of a film is labor. So, if I can replace a transcriptionist or an assistant producer or an archive producer with a machine, that’s going to cut down the budget. It also means a loss in jobs, which is not healthy for our community. And it also means a cut in quality.”
Basil Shadid (“Drag Becomes Him”) thinks AI will be a good thing for indie docu filmmakers without a streamer or network attached to their projects.
“As an independent filmmaker, like every other independent doc filmmaker, I wear 10 different hats when I am producing a film,” he says. “These AI tools might take three of those hats or make all 10 hats a little bit easier. But if I were a filmmaker with a fully funded film, I think it’s the filmmakers’ responsibility to hire humans. That is the thing that is going to be interesting to see play out in the future.”
That seems optimistic due to the fact that the majority of those working in the doc space, especially those on the low end of the totem pole, don’t have a union to protect them from a conglomerate like Netflix or Amazon to make sure or even fight for that to happen.
One thing that everyone can agree on is that AI is part of documentary’s future. Upholding the notion that documentaries represent truth, will be the genre’s biggest hurdle.
“I used to work at ABC News and my boss would constantly say, ‘If we lose the trust of the public, we are done,” says Porter. “And that is where we are right now with documentary. Right now is a really good inflection point for us to say, ‘what is a documentary?'”
