Second Take: ‘Megalopolis’ trailer AI fiasco speaks to larger issues with future of media

(Helen Sanders / Daily Bruin)
By Martin Sevcik
Nov. 3, 2024 9:31 a.m.
This post was updated Nov. 5 at 7:38 p.m.
Francis Ford Coppola’s latest film generated a media storm, even as its review scores slept with the fishes.
“Megalopolis,” a high-art science fiction Roman epic by “The Godfather” director Coppola, is floundering at the box office, grossing less than one-tenth of its $120 million budget since its Sept. 27 release. The film was initially written in 1983, spending the past 40 years as a production nightmare for Coppola, who repeatedly failed to find a compatible production studio for his vision and ultimately opened a winemaking business in 2010. The once-esteemed director returned in 2019 to spend that viticulture fortune on a film that has since put the “rotten” in Rotten Tomatoes – all while Coppola collects lawsuits from his alleged sexual impropriety on-set.
[Related: Second Take: Clothing reselling impairs fashion culture, prioritizing profit over expression]
With such a rocky production, it makes sense that a few things would fall through the cracks – such as good screenwriting, acting or visual effects. But one error from the film’s pre-release trailer, made by a marketing consultant who was fired because of the whole ordeal, is worth dwelling on as an example of the nature of truth in our evolving digital spaces.
After the film received negative reviews following its Cannes film festival premiere, a trailer emerged to highlight the similar initially negative reviews of Coppola’s other classic works. For example, critic Andrew Sarris of The Village Voice described “The Godfather” as “a sloppy self-indulgent movie,” while Vincent Canby of the New York Times described “Apocalypse Now” as “hollow at the core.” But a quick reread of those old reviews revealed none of the featured quotes were real – they had been fabricated for the trailer. Even worse, it soon came to light that the quotes had been generated by artificial intelligence, leading to the trailer being pulled and the responsible employee being terminated.
This situation is a comical cherry-on-top for a nightmarish production cycle, serving as another blow against a film that has had a tough life. But while this is currently a surreal, isolated incident of AI-generated slop in the film industry, it should not be treated as a fun novelty. It is an omen of things to come – and the lessons learned from “Megalopolis” need to be brought to the internet at large.
Consider the necessary thought process behind the inclusion of these quotes. The marketer supposed that something was true – Coppola’s films have historically reviewed poorly when first released – but was unwilling to find real evidence for their point. Instead, they used artificial intelligence to simulate that truth, to pretend as though they had found evidence. It is impossible to know the exact prompt used – did the marketer ask ChatGPT to find some real quotes or just to generate nonexistent quotes? – but using the oft-unreliable tool opens them up to falsehoods either way. With how well-known chatbots’ hallucinations have become in the past few years, there is no excuse for the marketer to not double check the bots’ work.
And yet it was approved anyway, presumably by a large team of people who edited, scored and managed the production of the trailer. This was not a subtle sleight-of-hand either – the trailer was immediately exposed as AI-generated slop once it hit the internet. So why did it manage to evade detection?
Because it is a trailer, meant to tell a story about the film’s ratings that would draw audiences to theaters. At no point in the trailer-making process did it matter if the quotes were true. Instead, it only mattered if they felt true.
In this case, it is a funny mistake. But the same logic is beginning to creep into other uses of AI-generated content with real-world consequences.
As “Megalopolis” screened in theaters across the country, Hurricane Helene hammered North Carolina and Florida. Extensive damage flattened homes and upturned lives as over 100 people died. In the wake of the hurricane, AI-generated, partisan images flooded social media, with false images of Donald Trump saving hurricane victims attaining viral success. In the midst of this media storm, Republican National Committeewoman for the Georgia GOP Amy Kremer posted an AI-generated image of a young girl escaping a flood. When she was called out for posting a fake image amid a real-world tragedy, she posted the following:
“There are people going through much worse than what is shown in this pic,” Kremer said in a reply on the social platform X. “So I’m leaving it because it is emblematic of the trauma and pain people are living through right now.”
In other words, it did not matter if the image was actually true. It only mattered if it told a good story – if it felt true.
[Related: Second Take: Chappell Roan is expressing valid concerns regarding celebrity-fan relationships]
The substitution of AI-generated slop in place of truth was outlandish in the trailer, but this approach becomes even more upsetting as it begins to cause real-world consequences amid real-world disasters. The film, a product of 40 years of dedication from Coppola, will likely have limited cultural impact – but its trailer, seemingly hastily made without care, will be seen as a shining example of AI-generated misinformation. It is an example of how, when fiction is easier to create than truth with new tools, the feeling of truth and truth itself will be treated as one and the same.
In one fell swoop, the “Megalopolis” trailer captured the fall of Rome, the final fall from grace of a respected director and the perils of misinformation of the modern digital age.