BUILDING A STRUCTURE: THE BRUTALIST and THE USE OF ARTIFICIAL INTELLIGENCE IN FILMMAKING

By Ray Wu

The Brutalist (2024), directed by Brady Corbet, is an ambitious VistaVision epic spanning architecture, immigration, class struggle, capitalism, Zionism, and addiction. The film follows László Tóth (Adrian Brody)—a Jewish-Hungarian architect and Holocaust survivor—who grapples with the labyrinth of postwar America, where power seems to loom over every aspect of his pursuit of agency.

The production is grand—so grand that it warranted a fifteen-minute intermission printed on chemical film. For Corbet, operating with a budget of a relatively low $10 million while casting actors like Adrian Brody and Guy Pearce, making the film could not have been easy. Consequently, he turned to artificial intelligence for assistance. In recent months, many have cast a spotlight on Corbet’s choice to incorporate AI, with certain applications sparking far-reaching conversations about the future of the medium in the creative realm.

AI-Assisted Dialogue Polish:

First, the less controversial bit: AI for dialogue fixes. Corbet and his team were transparent in their use of the audio tool Respeecher for some Hungarian dialogue refinements. According to Corbet, AI was only used “to refine certain vowels and letters for accuracy,” and “no English language was changed.” Still, the very involvement of AI-generated a huge backlash in the film community for reasons that remain unclear to me. It seems that everything creatively demanding about the dialogue was produced by a real person. In fact, manipulations to the dialogue, such as looping, editing, and mixing tracks, have existed for as long as films have had sound. Automated Dialogue Replacement, commonly known as ADR, has been widely in use since the 70s. In this particular case, the tool Respeecher was used as early as 2018 and is credited for synthesizing young Luke Skywalker’s voice in The Mandalorian. The involvement of AI, for The Brutalist, simply expedites the tedious process that ultimately wouldn’t have benefited the final product and would have delayed the release had Corbet adopted the “artisanal” and manual approach. 

Generative AI and Architectural Designs:

Conversely, the more provocative debate centers on the film’s Venice sequence set years after the main events: a public exhibition of Tóth’s brutalist buildings, some of which were conceptualized or partly derived from AI-powered imagery. Filmmaker Magazine notes that production designer Judy Becker and an architecture consultant “used Midjourney to create three Brutalist buildings quite quickly,” originally planned to be redrawn by a human illustrator. However, it remains unclear whether humans ultimately recreated the designs.

In this case, the AI footprint is far more nuanced than the previously mentioned “dialogue-gate.” Multiple sources have suggested varying degrees of AI remnants in The Brutalist’s final cut, making it difficult to parse what, if anything, constitutes misuse. One could argue that using artificial intelligence to generate preliminary concepts is akin to looking at other artists’ works for inspiration—so long as no purely AI-generated imagery appears in the film. But where do we draw the line? If these quick, algorithmically produced elements merely slip into the background, do we risk blurring the once-sacrosanct boundary between handcrafted cinema and expedient shortcuts? And if those bounds are repeatedly transgressed, might a filmmaker become comfortable featuring prominent AI visuals in the foreground, disguised by a layer of human refinement?

Moreover, some viewers, including my architecture professor, condemned the building designs in The Brutalist as “ugly,” but whether those opinions were shaped by their knowledge that AI was involved remains ambiguous. This speaks to a deeper tension: the understanding that, for many filmmakers and cinephiles, a film’s artistry is inextricably tied to its painstaking craftsmanship. Each of the 24 frames per second demands acute attention to detail. As exemplified by directors like Stanley Kubrick or Alfred Hitchcock, who famously supervised every inch of mise-en-scène, near-obsessive control can yield a rare cinematic magic. Whether it’s an out-of-focus, background architectural sketch or an elaborate set piece, the sum of details that often go unnoticed by a theatrical audience is also what subliminally seeps into their memory—capturing cinema’s unique ability to fabricate an escape from reality for hours. 

The Academy’s Response

The heightened controversy of AI assistance in film likely stems from the recent deluge of AI-generated images saturating search results, social media feeds, and a raft of online platforms. Much of this so-called “AI slop” is perceived as the lack of tangible human imprints or souls—an assessment that, in turn, has engendered a reflexive aversion to the technology in some filmgoing circles. The industry, too, has been justifiably rattled by studio executives’ unabashed discussions about AI’s potential to generate actors or backgrounds, from the SAG-AFTRA strike of 2023 to the continued inflaming concerns over job displacement.

Taken together, these factors prompted the Academy to establish stricter requirements for productions to disclose AI usage. While a voluntary disclosure form already exists, a formal, compulsory framework may soon be adopted to address rising demands for transparency. Still, it is worth noting that earlier precedents—such as digital de-aging in The Irishman or machine-learning techniques used to restore archival audio in documentary films—did not provoke the same outcry. The distinction may lie in the scope and immediacy with which modern generative AI can conjure entire illustrations or sequences, raising more complex questions about ethics and aesthetics.

As filmmakers continue to experiment, it has yet to be clarified where, exactly, ethical or creative boundaries regarding AI should be drawn. Given that new technologies often require a period of assimilation before stable norms emerge, outright condemnation might be premature. The real tension persists between an artist’s desire for total creative agency and the realities of an industry under constant pressure to innovate and economize. As the debate unfolds parallel to AI’s rapid evolution, directors, production designers, and industry regulators must grapple with the demanding endeavor of negotiating standards that safeguard cinema’s human essence without unduly stifling new approaches to filmmaking.

Whether this debate will ultimately yield a fresh golden age or undermine the intimate craft at the heart of filmmaking is still an open question. The allure of pure creative control, painstakingly executed frame by frame, has long defined cinema’s magic. Once that alchemy is mingled with rapid algorithmic generation, some viewers will undoubtedly continue to question if the sense of wonder might slowly give way to something less profound. Only time will tell whether AI, once fully integrated, will elevate the art form—or gradually steer it toward a mode of creation that sacrifices the soul for logistical shortcuts.