Can a computer write the next indie hit?

Algorithm-generated storytelling: the pros and cons of taking some of the risk out of filmmaking.

With horror feature Impossible Things – a screenplay co-composed by a computer – set to shoot in Canada in 2017, an implied question is rumbling: what role, if any, does artificial intelligence have for indie productions? And does it compromise the nature of creativity?

In the case of Impossible Things, the plot points in the story were suggested by an algorithm, after which a screenwriter (of the human variety) wrote the script. (While its name shares a potentially confusing similarity to Netflix hit Stranger Things, work on Impossible started well before the ’80s pastiche dropped on the SVOD). The computer program, developed over five years by Jack Zhang, founder of Kitchener, ON-based Greenlight Essentials, analyzes the plots of popular films, stripping out factors that might have artificially inflated its success (like big name stars, marketing budget). With this info, the computer comes up with a list of potential storylines and characters a movie could follow.

It then further profiles audiences, from the demographic and geographic makeup, to the potential enthusiasm for certain projects. Working off public and paid databases (the names of which Zhang did disclose), the AI can drill down into household income, purchase patterns and social behaviour, to further inform the content.

Impossible Things is not the only film tapping into computer-generated storytelling. Short film Sunspring (produced by California’s End Cue), released earlier this year, went the whole hog and used AI to write every word. The rather nonsensical nine-minute short was penned by an AI named Benjamin (created by filmmaker Oscar Sharp and NYU researcher Ross Goodwin) and received praise for its odd charm.

Seeing potential in Zhang’s platform, film financier Productivity Media (Born to Be Blue) partnered on both the horror film – acquiring the worldwide rights jointly with marketing and distribution firm Concourse Media – and as an adopter of Zhang’s technology to further its decision-making capabilities. With both sides considering the partnership to be mutually beneficial, no fee was exchanged.

One of the most attractive facets of this platform is in mitigating financial risk, says William Santor, CEO, Productivity Media.

“We saw so many uses for it for producers, sales companies and lenders,” he says of the tech’s applications at all levels of the production chain.

The idea is that before filming (or even writing) begins, the program can provide a specific profile of what the audience for the project looks like and what stories will resonate most.

“As a lender, we saw a chance to de-risk part of our lending portfolio using data,” says Santor.

Understanding your audience before you are too far down the rabbit hole with your project is the goal here. And at the indie level in particular, where financial resources are more scarce, the onus to maximize every dollar is even greater, argues Santor.

“If you have the ability to listen to your audience before you spend the money to create something, I think it’s prudent and responsible,” he says. By “beginning with the end in mind,” content creation can be enhanced because creators have a better understanding of their audience and produce laser-targeted work.

In terms of risk mitigation, the rules of the game have changed now that Netflix and Amazon are mining vast quantities of user-behaviour info before deciding where to allocate their resources. This, says Santor, is the new norm.

With Netflix’s House of Cards, for example, user data showed bringing together The Social Network director David Fincher with Kevin Spacey in a remake of a successful political thriller would be the recipe for a hit. It was right.

And while it is still early days for Productivity using this platform, Santor says the validation the company has already seen has been “exceptionally compelling.”

But not everyone sees it this way. Some of the creative community consider the concept of a computer dictating the direction of a story as an affront to creativity.

“When you make safe bets you get safe results,” says filmmaker Jeremy Lalonde (How to Plan an Orgy in a Small Town). “In theory, [this type of software is] creating a scenario where people are just doing safe things and pandering to their audience.” And what happens if the data all points to the same direction? Will we just get more of the same?

Engineering storylines to appeal to certain audience segments has long existed under one guise or another. The key is knowing when a storyteller’s instinct should override what an algorithm suggests, Lalonde says.

To some degree, the industry’s stance on creating content in this way might already be softening, says Greenlight’s Zhang. When he first introduced his program at TIFF 2014, the reaction he got from producers was arm’s length, though when he returned this year the reception was far warmer.

The end game is not to replace writers with computers, he stresses, but align the interests of creators more closely with audiences.

No, robots and algorithms aren’t replacing us all. But they might be able to help us out.

Impossible Things is set to begin filming in Ontario early next year.

This article originally appeared in 
Playback’s Winter 2017 issue.

Image: Shutterstock