I’ve never been someone who gets nervous about the perils of advanced artificial intelligence. Even with brilliant minds like Stephen Hawking warning us that uncontrolled AI could lead to “the end of the human race,” I’ve always maintained that humans will be able to handle whatever sentience springs forth from future technological developments. I feel like the possibility of a “murderously intelligent, sensually self-programmed non-being” like Proteus IV from Demon Seed (1979) is about as likely as an invasion of killer alien clowns (from, well, you know).
So from my perspective, developing artificial intelligence in robots is perfectly fine. I’m all for it. The sooner the better. But what about the day when AI is used to write scripts for movies? I find that absolutely terrifying. The scariest part? That day has already arrived.
Recently, a film titled Impossible Things was successfully backed on Kickstarter. On its campaign page, Impossible Things is touted as being the “world’s first feature length film co-written by artificial intelligence.” Here’s a look at their Kickstarter video including a teaser trailer for the film itself.
The Kickstarter page explains that AI was used to create the premise for the movie as well as its major plot points. The software uses data gathered by breaking down the plots of previous films and matching that data to “audience taste.” This seems a little ambiguous, but from the best of my understanding, “taste” is used in reference to box office performance.
Greenlight Essentials is the software company behind the program used to create the foundation for the script of Impossible Things, and the company’s goal is to bring big data into the film industry. As stated on their own web site, their intention is to allow filmmakers to “explore and discover repeatable patterns from decades of film data” without the hassle of “watching decades of films or gathering past audience reactions.” The goal, it seems, is to allow writers and producers to see what has worked in the past in order to better tailor their films to a specific audience, thereby increasing its chances of monetary success. If this sounds like an unusually cold and detached way of making a movie to you, then I’m right there with you.
Even so, I can see where this technology would be very attractive for film studios. You know, the studios that like to put money into remakes, sequels, and other safe bets when they see that, for example, Poltergeist (2015) made over three times the money that It Follows (2015) did domestically. Now, I understand that the budgets for these two films are vastly different which affects the relative “success” of each title, but I chose these movies because they are prime examples of the movie-going public not necessarily spending more money on good movies as opposed to bad movies. According to Rotten Tomatoes, the Poltergeist remake was panned by critics and audiences alike, yet It Follows was beloved by critics and generally liked by audiences.
I feel that originality and integrity of vision are two of the most important things when it comes to making movies. The minute a filmmaker starts to compromise their vision in order to win over a wider audience is the minute their movie starts to become watered down. I don’t want filmmakers to pander, I want them challenge an audience. Some of the greatest and most influential movies in history were considered bombs. Freaks (1932) was a box office failure and essentially destroyed the career of Tod Browning, the man who directed Bela Lugosi in Dracula (1931). Test screenings (the traditional way of discerning an audience’s tastes) for Freaks were disastrous and led to the movie being cut down from around 90 minutes to around 60 minutes. Today, Freaks is regarded as a hugely influential and important film. If Tod Browning had access to software from the start of production that spit out a plot and setting according to audience’s spending patterns at the time, Freaks would almost certainly have never been made.
Perhaps I’m being a bit dramatic. There will always be people out there creating movies to express their vision rather than to cater to an audience, but I feel that the AI being developed by Greenlight Essentials has the very real possibility of simply validating the tendency of big studios to put money into safe projects rather than taking a chance on something new. Would Greenlight Essentials’ program have dissuaded Artisan Entertainment from picking up and distributing a genre changing movie like The Blair Witch Project (1999)? Like it or not, that movie changed the landscape of the horror movie industry and did quite well at the box office, but was there a precedence for it that a computer program could have detected? I feel like the best movies are often the ones that come out of nowhere. They defy expectations.
Do we really need a program to help us write horror movies that will make money? I think the formula is pretty simple: blood, nudity (though this is sometimes negotiable), and a low budget. If you keep the budget low enough, then the return on investment is much easier. That doesn’t mean the movie will be good, but mining plot points from previous movies doesn’t mean it will be good either. There are enough derivative horror movies out there already. Just take a look at the Syfy Channel.
With all that said though, I’ll keep an open mind when Impossible Things finally comes out. It might be good. After all, most stories have already been told at this point, so it’s how you tell it that really matters. If the production team can breathe life into the cold, lifeless way in which the plot was constructed, then maybe it will help alleviate some of my fears. Maybe it will be time to roll out the electric red carpet and welcome our new AI movie-writing overlords.
References & Links
 – http://www.bbc.co.uk/news/technology-30290540 (“Stephen Hawking Warns Artificial Intelligence Could End Mankind”)
 – https://www.youtube.com/watch?v=H6O1NRs-YuU (Demon Seed trailer)
 – https://www.youtube.com/watch?v=cVQ3AGzeB_0 (Killer Klowns from Outer Space trailer)
 – https://www.kickstarter.com/projects/970954017/impossible-things-worlds-first-ai-co-written-featu/ (Impossible Things Kickstarter page)
 – http://www.glessentials.com/ (Greenlight Essentials web site)
 – http://www.the-numbers.com/market/2015/genre/Horror (“Box Office Performance for Horror Movies in 2015”)