On the Topic of Faking It: Machine Learning and Authentic Acting in Modern Film

Knowledge hub

Adrien Brody as architect László Toth in Brady Corbet’s The Brutalist (2024). Image courtesy of A24.

On the Topic of Faking It: Machine Learning and Authentic Acting in Modern Film

Written by Callum Stewart

Brady Corbet’s immigration epic The Brutalist has been sweeping the 2025 awards season. Critics have lauded the film, describing it as an “uncompromising cinematic statement ”. But in early January, film website Red Shark News published an interview with editor Dávid Jancsó, where he discussed using AI voice program Respeecher to correct the Hungarian accents of leads Adrien Brody and Felicity Jones. This sparked an online firestorm, with many retracting their praise and accusations of an ‘artificial’ performance abound. But if use of these tools makes a performance artificial, what constitutes a real performance? Where does advanced CGI fit in? How do we define ‘faking it’?

“AI” has become one of the more insidious buzzwords of our time, seen on everything from smartphones to rice cookers. Yet the way it’s used is often inaccurate. What tools like ChatGPT, Respeecher, and Midjourney employ is machine learning (ML). These are systems trained on a dataset that then regurgitate content based on their training – in the words of Ben Affleck, they “write excellent imitative verse…cross-pollinating things that exist. Nothing new is created.”

In the case of The Brutalist, Jancsó – whose mother tongue is Hungarian – trained Respeecher with his own voice. It’s understandable why these varied tools have all fallen under the same “AI” umbrella; it’s a shiny yet familiar term, with years of sci-fi hype around it. ML can be applied for a wide variety of applications beyond generative content; but by describing all these tools as “AI”, we find ourselves in murky waters.

Actor Guy Henry (right) before and after being deepfaked to resemble the late Peter Cushing (left) in Rogue One: A Star Wars Story (2016). Image courtesy of Disney.

To some moviegoers, AI is a dirty word – an omen of money-grubbing Hollywood necromancers, intent on resurrecting dead stars and stealing opportunities from working actors . SAG-AFTRA, the world’s largest union for performers, struck in 2023 to secure rights around the use of ‘digital actors’. In March 2024, the directors of indie horror flick Late Night with the Devil mentioned the use of AI to generate three interstitials for the film’s talk-show set design. The public response was volatile, with one review on Letterbox saying: There’s AI all over this…It actually feels insulting…like the filmmakers don’t give a shit.” Another review warned of setting a precedent, noting that “ Complacency in accepting AI now is complacency for AI in the future – a very bleak future”. Generative tools also pose ethical concerns, particularly around copyright and their environmental cost. A thousand image generations from Stable Diffusion XL produces approximately as much CO2 as driving 6kms in a petrol car – multiply that by an estimated 15 billion generations a year and the issue is apparent.

Altering a performance through special effects is nothing new – consider motion captured CGI, or the many musicals that autotune their lead’s voices. Not only is the performance still the product of an actor, but the alterations made are from the skill of an editor – not the “imitative verse” of a generative model. Speaking about his use of Respeecher,Jancsó stated that “you can do this in ProTools yourself….There’s nothing in the film using AI that hasn’t been done before. It just makes the process a lot faster”.Is The Brutalist’s voice editing any different from The Irishman’s AI-assisted digital de-ageing , a process that Scorcese used explicitly to enhance his actors’ natural performances? Allegations of ‘digital doping’ seem excessive – used tastefully, there’s a valid place for these tools.

Robert de Niro before and after the digital de-ageing process in The Irishman (2019). Image courtesy of Netflix.

Questions of authenticity aside, many consider the mainstream adoption of these tools to be inevitable (as with anything that increases productivity). Attitudes toward them will likely soften over time, either from acceptance or resignation. Is there something creatively lost in the endless climb to greater efficiency? Probably. Generative AI remains a very real threat to working actors, and regulation will be critical to protecting both the artistic and financial livelihoods of filmmakers. Famously crotchety director Paul Schrader recently sung the praises of ChatGPT – I suspect that parts of the filmgoing public will not share his enthusiasm.

Further reading:

"AI voice technology used in The Brutalist is nothing new – the backlash is about transparency", Dominic Lees, The Conversation, 2025

"Explained: Generative AI's Environmental Impact", Adam Zewe, MIT News, 2025

About the author

Callum Stewart

Callum Stewart is the Partnerships & Industry Events Coordinator at the National Communication Museum (NCM), where he leads the Industry Events program, connecting experts across various fields. With a background in fundraising at the Abbotsford Convent Foundation and curating for the Human Rights Art & Film Festival, Callum brings extensive experience in the cultural sector. He has also been long involved with Melbourne's film festival scene, having been a member of the Melbourne International Film Festival’s Shorts Panel for five years. With a deep passion for film and a fascination with technology, he is particularly interested in the intersection of the two, exploring how advancements like AI are shaping the future of the industry and storytelling.

According to Callum, most of his psychological profile can be found on his Letterboxd account .