by HyperAI
As of last Sunday, the global box office revenue of "Avengers 3" has reached 1.686 billion US dollars, currently ranking fourth in the global box office history.Creating a film like "Avengers 3" that is both a box office success and a critical success has always been the dream of filmmakers.
Recently, three technology companies have developed three AI programs that can predict future movie box office, and they claim that the accuracy rate has reached 80%.
However, can a product like movies, which is highly creative and has market randomness, really be perfectly predicted by AI?
You can know the box office before it’s released
Almost all well-known film companies will release new movies in summer and autumn, and will also hire professional market analysts to analyze public tastes during the preparation period of the movies.
AI predictions are actually the same as most market researchers, who analyze consumer behavior based on historical box office data, but it is obvious that machines are definitely better at processing data than humans.
There is an AI startup called Vault from Israel. Since 2015, they have been committed to using neural network algorithms to analyze box office data from the past 30 years, nearly 400,000 scripts, movie budgets, and overall audience numbers to estimate the box office of new movies in their first week of release. Its founder and CEO David Stiff said that their AI system can now predict movie box office with an accuracy of about 75%, which is very close to the actual box office.
At the beginning of the company's establishment, he also said that the program could predict movie box office with an accuracy of 65% to 70%.
Another Boston-based technology company called Pilot, inspired by Netflix's "What to Watch Next" feature, took a completely different approach to forecasting. Based on a large amount of movie information generated over the past 30 years, they combined multiple variables such as actors, directors, screenwriters, budgets, and plot summaries to predict the box office in the first week after a movie is released.
The company claims to be able to predict 70% of results accurately, and 80% of box office revenue after the first trailer of a new movie is released. It’s really hard to imagine that this accuracy is achieved by an application that has never seen a movie.
Can AI really predict movie box office?
However, filmmaking itself is a very creative thing. If the creation of a film is tied to the conclusions drawn from studying historical data, and the product of this creativity is turned into an assembly line product, will the film still have such great appeal?
In addition, whether people decide to watch a movie does not entirely depend on the historical success of movies of the same type. More decisive factors lie in the attractiveness of the script, the sophistication of the production, and the expressiveness of the actors themselves.
Before Marvel launched superhero movies, no one thought that a movie adapted from a comic book with such little attention and starring a newcomer would be so successful. If the predictions made by the three AI startups based on historical data were followed, Marvel's first film "Iron Man 1" would be the film that should not have been made, because there is no precedent for the same type of film in history, and Robert Downey Jr. would not have become one of the highest-grossing actors in history.
If Legendary Pictures had adopted its precise marketing strategy, Iron Man 1 would not have appeared, because its audience was limited to Marvel's hardcore fans, and no other audience could be found. Of course, it cannot be denied that the conclusions drawn based on historical data are meaningless for future film production, but if it is used as a basis for creation, it is likely to miss the next Iron Man.
This is like the professional sports clubs that rely most on data. To win the game, they still have to rely on the good chemistry of the entire team, rather than relying solely on summarizing historical laws.
There is no denying the positive impact of AI on filmmaking, but good movies should still be made by humans.
After all, the movie is a story about people, not machines, otherwise it seems that AI will sooner or later bring us all into "The Truman Show".