[Editor’s Note: This piece is written by an individual with a long background in the tech community who prefers to remain anonymous.]
No one wants to watch a bad movie. But what makes a movie “bad,” exactly?
If you’re the audience, a “bad movie” is one that you don’t like. Now, there’s a calculation that’s impossible to get wrong. Critics be damned—we like what we like! Speaking of critics, they usually offer more technical answers to the question of what’s bad in film. (Spoiler alert: if the critic doesn’t like the movie, it’s bad.)
But what if you’re not in the audience at all? What if you’re on the other end: the producer, the studio? From your perspective, the bad movies are the ones that lose money. Not surprisingly then, bottom line thinking is driving the latest trend in Hollywood: Let artificial intelligence software (AI) help analyze and possibly predict hit movies.
Will it work? There’s a story in that…
Last week, Warner Brothers announced that it is now using the Cinelytic platform to “guide decision making at the green light stage.” (Hollywood Reporter) To hear them tell it, the future is wow. But dig further into the buzz and discover that Warner Brothers is merely using the platform as one signal among others, a source of information to consider about which movies to green light.
The studio’s interest is understandable. Despite its visceral appeal, gut instinct has not served studios well:“In a sample of 50 films, including Hereditary, Ready Player One, and A Quiet Place, just under half made a profit, giving the industry a 44 percent accuracy rate.” (Verge) Verge also reports that a rival software system known as ScriptBook outdid the gut instinct of studios, predicting hits with 86% accuracy.
Whether this is “AI” is debatable however. The Cinelytic software analyzes factors or variables thought to contribute to a film’s success such as A-list actors, choice of director, and film elements like genre, plot, and style. In this regard, Cinelytic’s “AI” package helps streamline and extend the back-of-the-napkin calculations that inevitably figure into Warner Brothers funding decisions. Given the same information, one wonders if similar predictions could be made via the inspired use of macros in Microsoft Excel.
But the devil, as always, lurks in the details. Studio execs greenlight films before they are made—before things can go wrong. But the performance figures touted by the marketing departments of software firms are predictions made after the fact, analyzing known film successes and failures. We are not comparing apples with apples here.
As critics point out, Cinelytic’s software, and that of other tech startups like ScriptBook, does not add new thinking tools to our arsenal. It works best when relying on factors that are obvious to anyone who considers them. For example, the proposition that, in general, adding A-list actors will boost revenue is not difficult to grasp. The software usefully gathers into one place variables that we already know can be helpful. Yet a finer-grained analysis of factors that are in the end still guesswork can be dangerous—very dangerous—to decision-makers, particularly when the software is marketed as truly “intelligent.”
The key problem with viewing such software as superior to “gut” instinct is that in neither case is past performance a guarantee of future success. But , a perverse incentive can arise from using expensive AI platforms for future prediction. We may come to believe that we can game the system by actually predicting audience reception of creative content like movies.
Patterns exist, of course. But hits are often surprises, which by definition are not predictable from prior performance (that’s why it’s a surprise). This is part and parcel of the film industry; there is an unavoidable volatility in making motion pictures released to large audiences—it’s part of why we love cinema. Unpredictability is baked in, and functions as a guard against using pat formulas, a strategy that inevitably threatens creativity. Studios—producers, directors, and actors—strive to bring something new into the world. A hedge against such creativity drying up is acknowledging the threat of money loss and recognizing the inherent risks involved. Warner Brothers—and other studios—beware.
To be sure, analytics can be an important tool in a studio’s toolkit, but (ironically) only if the predictive element central to the marketing pitch of companies like Cinelytic is not taken too seriously. Huh? That’s right. We don’t want hit movies to become predictable—that’s why we pay the actors, studios, script writers and everyone else so much money.
It’s not a real worry, anyway. As in other industries with unavoidable volatility (say, the stock market), methods for taming risk only work for a while. They are, in effect, pseudo-methods that create the illusion of calculating risk (called “risk assessment”), and projecting future earnings—as Warner Brothers hopes to do with Cinelytic. They spark informed discussions. They don’t work if taken too seriously.
The bottom line: Hardwired into the nature of things is the fact that we can’t predict surprises. In a risky business like filmmaking, you can release a string of duds and then Joker, which pays all the bills and then some. Bur what if we did know what would pack the house in the 2020s? Everyone would soon be doing it and our great idea would be standard fare. AI will not change that pattern either. It can’t.
AI does pose at least one threat to filmmaking. It could intensify the very tone-deafness that studios hope it can fix: Too much reliance on ever more finely grained analysis of the patterns in past data could blind decision makers to the real risks, volatility, and opportunities in the future. That’s a recipe for losing money and inflicting “Oh, not that again!” on audiences. Let’s hope movie execs—historically a “gut feel” crowd for better or worse—retain their love of film, and use AI software as an aid to what will always be a human exploration and celebration of storytelling.
Note: Mind Matters News’s film critic, Adam Nieri, also weighs in: AI goes to Hollywood: Cinelytic’s AI does is what every AI does at its most basic level. It collects and sifts large amounts of data to find patterns that can be used to make predictions. AI excels at pattern recognition because that is its only purpose. Humans aren’t so linear or predictable.
Further Analysis at Mind Matters News:
The Golden Age of the Web?— A Dissent What happened to the collaborative culture, decentralized markets, and wisdom of crowds that bestsellers prophesied fifteen years ago?
We built the power big social media have over us Click by click, and the machines learned the patterns. Now we aren’t sure who is in charge
Futurism doesn’t learn from past experience. Technological success stories cannot be extrapolated into an indefinite future