When news came last month that Warner Bros. was using a startup's artificial intelligence program to "make decisions about content," people understandably freaked out.
“Hollywood is replacing artists with AI. The future is bleak, ”says the headline. "Warner Bros. wants to use AI to predict the box office, so start the terminator jokes," says another article.
On Twitter, chess master Garry Kasparov, who once lost a match against IBM's Deep Blue computer, called "the worst idea since they taught them how to play chess!"
But the people reading the headlines may have overreacted. Cincttic, the company that Warner hired, does not have an A.I. Movie Exec reads through a bunch of scripts and chooses which will be the next wildcard or parasite. Instead, it is a “project management system” that uses a machine learning algorithm based on historical data and current industry trends to produce a series of sales forecasts in real time. These are based on a number of variables, including cast and release schedule (date, market, theater versus streaming, etc.), as well as features such as genre or whether the film is a sequel or adaptation.
With some accuracy, executives and financiers can see the potential performance of a film and make an informed decision about how to proceed. That can mean anything from instantly rejecting a project to big bets with more actors on the A-list or a bigger theatrical release.
To give Fortune a better understanding of how Cicinetic works, CEO and co-founder Tobias Queisser demonstrated the product to us and answered our biggest questions about service, including what he can and cannot do, and how he can help make independent films and why it could increase diversity in Hollywood projects.
Cinestic co-founder Tobias Queisser has a background in both film production and finance. Courtesy of Crinetic
Why start Cincttic?
Catelytic was founded in 2013 by Queisser – who has a background in both film production and finance – and Dev Sen, who spent 15 years at NASA to develop risk assessment software to analyze the design of launchers. They founded the company after realizing that the film industry hadn't kept pace with the technological advances in data science and basic information exchange used in other industries, and later hired COO Christian Monti, a former studio manager who international sales were at Paramount and Warner. Her customers include STX Entertainment, Sony and most recently Warner's international division.
“The film industry is already a very risky business. The main reason is that content has to be funded in advance, ”says Queisser. "You have a script, you wrap it up with an actor or director, and then you come to a point where you have to finance it. You have to understand, is this package worth $ 10 million, $ 50 million, $ 100 million more? "
And while filmmakers could see the end dates of a project over the next few years and wonder what went wrong or right with their investment, the Cicinetic team developed their system to find out this value at an early stage when most of the Work is done through gut instinct and inefficient data collection in isolated departments. The best thing in Queisser's eyes is that they not only figured out how to pre-evaluate a project, but also sped up the whole process.
"The main advantage is that the workflow is optimized," he says. "And processes that took two days and did a thorough analysis of your project now take 20 minutes."
An example of a Cicinetic dashboard for the hit "Once Upon a Time … in Hollywood". Courtesy of Cincttic
What data is used?
With numbers and information from Nielsen, Studio System from Gracenote, Variety Insight and other sources, Cicinetic examines box office revenue and film performance on physical releases, SVOD, VOD, television, etc. in global and specific markets – all of which would add up to return on investment.
Movies are also analyzed based on a number of attributes, including key talent, genre, rating, sequel, customization / restart, key issues, country of production, distribution and number of cinemas. There is also a database of 550,000 actors, authors, directors and producers, which tracks their appearances in various films and markets and provides additional information about the importance they had for the project. For actors, this can mean writing down whether they did the first or second payroll in an action film or a lead role in an animated film, which helps them achieve economic talent. With a short list of potential actors, a producer could then enter different combinations of casting configurations to see what works best in Europe and Asia compared to North America, for example.
To make things more accurate, Cicinetic only uses the performance of the most recent films for its predictions and not historical data. “The film industry is changing very quickly, so you really want to focus on the last three to five years. There is no point in watching films from the 70s and 80s, ”says Queisser. "Streaming did not exist 10 years ago. DVD was a major source of revenue in 2002 and 2004. It is important to be as up to date as possible."
Cinetic predicted that the 2019 Hellboy would bring in $ 23.2 million at the box office at home. It turned out to be close to the film's actual $ 21.9 million figure. Summit Entertainment – courtesy of the Everett Collection
Where does the human element come into play?
When a user enters all the attributes of a film, they get predictions about national and international revenue and the confidence level of the algorithm in each projection, based on probability and risk-weighted scenarios. The mean confidence level, or “base case,” as Queisser calls it, is the most likely outcome, the safest bet in the group. As an early example, Cincttic used the 2019 Hellboy, with the middle projection being a domestic box office of $ 23.2 million, which turned out to be close to the film's actual number of $ 21.9 million.
This does not mean that the system predicted the film would be below average quality or that it was stupid for Lionsgate to spend $ 50 million on it. "Because of the technology we use – machine learning, A.I., and pattern recognition – you can't reverse engineer why that was said," says Queisser. "When we knew the project, it basically said," This kind of intellectual property with this talent may not work as expected. "This is exactly what our system does. We are unable to predict hits or flops."
He points out that Cincttic almost also found out the domestic footage of several other films from 2019, including What Men Want ($ 56 million forecast vs. $ 54.6 million actually), Rocketman ($ 91.7 million -Dollar vs. $ 96.4 million) and Once Upon a Time in … Hollywood ($ 126.1 million vs. $ 142.5 million). These were not a judgment of how good the films would be, just a distillation and analysis of the data.
"We don't think an algorithm can run a script and understand whether the story is good or connecting with people," says Queisser. "At the moment A.I. It's about cracking numbers and recognizing patterns, and we need to use this technology in the most pragmatic way at the interface between creativity and data. "
For industry decision makers, this means looking at the data and finding out where to take risks. A manager can view Cicinetic's analysis and use one of the projections that the system is less likely to trust if they really believe in a movie.
"The producer or manager can say," This is one of the best scripts I've read. I worked with this producer and the director is great. I am ready to give the green light or to work on this project if it is balanced in the (long-term) area, ”says Queisser. "We give them the opportunity to overlay their creative understanding of the project."
"Crazy Rich Asians" with Awkwafina (left) and Constance Wu (right) was a mega hit in 2018. Sanja Bucko – Warner Bros. Pictures / courtesy of the Everett Collection
How can it lead to more originality and diversity in the film?
Queisser says that a common misperception of his A.I.-based system is that it is predicted that "safe" films are the way to go, rather than projects that are more original or have different actors and creators. In his view, Cincttic does the opposite by including success stories like Moonlight, Crazy Rich Asians and Parasite.
"Only 4% of independently produced films actually make up for their investment, and every step in the value chain is done by a different company," he says. "These companies may have very different business models and are often not very well aligned, so this industry would actually be more sustainable because of the more objective use of information."
In terms of diversity, he says that Cicinetic's use of sales information over the past three to five years has led to the effective eradication of long-held prejudices that films by women or with different actors will not be successful. "Data doesn't discriminate," he says. “The latest data is very positive for different talents, especially in the past three years. We can show that these projects actually work. "
How does the system take surprise hits and flops into account?
In short, that's not the case. However, a manager could use real-time data to adjust their financial expectations and publishing strategy, based, for example, on early audience reactions or reports on how well a production is going. According to Queisser, Cincttic made predictions about Joker before it was finished, and the most likely result was that the film would debut in 3,600 cinemas and make $ 130 million. But when the Toronto International Film Festival came in early and critics were optimistic, Cincttic repeated the film's numbers with 4,300 cinemas released, increasing it to $ 270 million.
Hypothetically, knowing such a gust of wind could result in a studio giving the green light to another film that is a riskier investment. On the other hand, if the film had dolittle-like production problems and negative enthusiasm, the studio could have cut its advertising budget to minimize losses and done some safer projects, like a player taking a few dollars off the table after he got one bad hand.
“You want to take a pragmatic and methodical approach. You want to make sure that the baseline scenario is that you get your money back or make a small profit, ”says Queisser. "They hope for an outlier – this is the optimal scenario."
Other essential stories from Fortune:
– Breaking Bad ended six years ago – but Better Call Saul and merchandise are gaining momentum
– How the cult rock band Guided by Voices from the 90s flourished despite the chances of the industry
– With Sonic the Hedgehog at the box office, Paramount catches "lightning in a bottle"
– Rob McElhenney on researching misogyny and censorship while playing with Mythic Quest: Raven & # 39; s Banquet
– Contagion writers and scientific advisors reflect the newly discovered relevance of the film in the coronavirus crisis
Follow Fortune on Flipboard to stay up to date with the latest news and analysis.