Last Friday, Bloomberg reported that OpenAI, the artificial intelligence company that made ChatGPT, has scheduled meetings with “studios, media executives and talent agencies” in the hopes of pushing the industry toward integrating the use of AI tools into their film and television productions — particularly a new AI video generator that the company has developed. This report coincided with the release of the new independent feature “Late Night with the Devil,” a film that received good reviews at SXSW earlier this month but faced some backlash online when viewers discovered that it contained a handful of AI-generated images. Though the filmmakers defended their decision — brushing away the criticism because of the relatively slight usage — it has sparked plenty of criticism from those who believe we could be on a slippery slope to below- and above-the-line jobs automated by much cheaper AI tools.
Maybe you don’t care about filmmakers using generative AI in their films. But you might want to ask yourself why you don’t care. Do you think of the film and television you watch as art or as product — something to mindlessly consume while you do other tasks or just a way to pass the time? If it’s the latter, then of course you don’t care! You don’t think about how the media you consume gets made or who makes it. If it can look like a person made it, that’s good enough for you. Your consumption habits are already determined by a complex algorithm. Why can’t the things an algorithm tells you you might enjoy also be produced algorithmically? Let’s say that generative AI begins to dominate the industry. One would think that, eventually, people will get tired of the same old repurposed shlock and, in an attempt to regain mass profitability, the industry will adapt by dispensing with AI-driven projects and back to human stories told by people with new and exciting ideas. But I’m not sure that would happen. Audiences have been trained not to expand their horizons and that sticking with what you know is ideal and something to be proud of. It’s not, and it won’t be more so when ChatGPT is writing every episode of your new favorite television show.
Some people think generative AI is righteous, that it can be a cheap (or maybe even free) tool for independent artists to add “production value” that they would never be able to afford otherwise. Indeed, unimaginative, talentless hacks can now “make” images they would want to see in movies, but those images will always feel hollow. They give the surface-level appearance of cinema without any of the creative decision-making that makes film so exciting. It can only be as deep as the brief prompt given to the technology. There is no longer a question of why a choice was made because there was no choice. AI doesn’t make choices. It works on probabilities based on test sets to spit out what an algorithm tells it is most similar to the prompt fed into the system. It makes the ability to conjure images more accessible, but in the process, it dilutes the power of those images. It places more value on efficiency than personality, and the medium does not need that.
This whole debacle has got me thinking more about cinema as an art form, whether a medium that (at least in America) is becoming almost exclusively consumer driven even at the independent level can even be considered art, and what art is at its core. If art is the result of human creative expression, then can a film that is primarily the work of generative AI be considered art? What generative AI spits out is not human creative expression, but it is built on massive historical datasets of it — and, it should be noted, the creators of that data are not compensated for it. If that dataset should become overpopulated by AI-generated data, then eventually a feedback loop creates a scenario where we are seeing several copies of copies of the work of artists, but the humanity is far enough away that we can pretty accurately describe it as inhuman.
What of films like “Late Night with the Devil” that make minimal use of AI-generated images? Are we to discount the totality of human creativity that did go into it because some parts of it weren’t made by people? It’s tough to say, particularly considering the below-the-line craftspeople who have poured their hearts and souls into the project, but one or two people made the final decision to use AI elsewhere in the film. At the same time, we can’t let small things like that slide because the concessions to generative AI that we allow in our films will only grow until the technology has taken over most of the production process. Cultural criticism is not in a state to properly sway public opinion on this. We are experiencing a culture in which philistines and anti-intellectuals are the primary influences on taste and consumption patterns. Call me pretentious all you want, but I’d much rather stand in defense of art and artists than whatever OpenAI is cooking up for the major Hollywood studios. If we don’t take a stand now, it’s only a matter of time before it infects every other art form.
If the entertainment industry wants to continue its flirtation with these inhuman technologies, it is going down a very dark path. An interstitial here becomes a fully generated shot. A fully generated shot becomes a generated camera pan. A generated camera pan becomes generated environments, which become generated characters, which become scripts, and so on. This is not like computer-generated imagery — which, while sometimes fully digital, is still the work of human visual effects artists, and design decisions run through directors and production designers. It takes the personal element out of the equation. It takes human creativity out of the equation. It is not art. It is highly processed slop. It doesn’t matter how real it looks. If you would rather watch that than a poorly shot, poorly written, poorly edited, poorly acted, no-budget film made by actual people, this medium is not for you.
