The Verge reports that Storied film company Warner Bros. has signed a deal with Cinelytic, an LA startup that uses machine learning to predict film success. A story from The Hollywood Reporter claims that Warner Bros. will use Cinelytic’s algorithms “to guide decision-making at the greenlight stage,” but a source at the studio told The Verge that the software would only be used to help with marketing and distribution decisions made by Warner Bros. Pictures International.
In an interview with THR, Cinelytic’s CEO Tobias Queisser stressed that AI was only an assistive tool. “Artificial intelligence sounds scary. But right now, an AI cannot make any creative decisions,” Queisser told the publication. “What it is good at is crunching numbers and breaking down huge data sets and showing patterns that would not be visible to humans. But for creative decision-making, you still need experience and gut instinct.”
Regardless of what Cinelytic’s technology is being used for, the deal is a step forward for Hollywood’s slow embrace of machine learning. As The Verge reported last year, Cinelytic is just one of a new crop of startups leveraging AI to forecast film performance, but the film world has historically been skeptical about their ability.
Andrea Scarso, a film investor and Cinelytic customer, told The Verge that the startup’s software hadn’t ever changed his mind, but “opens up a conversation about different approaches.” Said Scarso: “You can see how, sometimes, just one or two different elements around the same project could have a massive impact on the commercial performance.” Cinelytic’s software lets customers play fantasy football with films. Users can model a pitch; inputting genre, budget, actors, and so on, and then see what happens when they tweak individual elements. Does replacing Tom Cruise with Keanu Reeves get better engagement with under-25s? Does it increase box office revenue in Europe? And so on.
Many AI experts are skeptical about the ability of algorithms to make predictions in a field as messy as filmmaking. Because machine learning applications are trained on historical data they tend to be conservative, focusing on patterns that led to past successes rather than predicting what will excite future audiences. Scientific studies also suggest algorithms only produce limited predictive gains, often repeating obvious insights (like “Scarlett Johansson is a bankable film star”) that can be discovered without AI.
But for those backing machine learning in filmmaking, the benefit is simply that such tools produce uncomplicated analysis faster than humans can. This can be especially useful at film festivals, notes THR, when studios can be forced into bidding wars for distribution rights, and have only a few hours to decide how much a film might be worth. “We make tough decisions every day that affect what — and how — we produce and deliver films to theaters around the world, and the more precise our data is, the better we will be able to engage our audiences,” Warner Bros.’ senior vice president of distribution, Tonis Kiis, told THR.
—
Photo Credit: metamorworks / Shutterstock.com