Sunday May 07, 2017
Precogs need not apply: can algorithms predict crime?
The new film Pre-Crime -- which recently premiered at Hot Docs -- shines a skeptical light on predictive policing methods. Sci-fi author Philip K. Dick coined the term "pre crime" in a 1956 short story, and inspired Steven Spielberg's dystopic 2002 film Minority Report.
What is pre-crime? "It's the idea of preventing crime before it happens," explains co-director Matthias Heeder. He points out that police don't like the term "pre crime" because it evokes The Minority Report, with which they don't identify.
So how are police actually using predictive algorithms to track potential "high risk" offenders? Andrew Ferguson is University of District of Columbia law professor who appears in the film. "In Chicago right now, an algorithm is identifying the top 1500 people who they believe are the most at risk of being violent or to be the victims of violence," Andrew says.
The data that determines who's on the list, is not publicly available, but critics point out that poor neighbourhoods and minority groups are often the biggest targets for predictive algorithms. "There is a potential for bias," Andrew adds. "You might create a self-fulfilling prophecy that if the police go to a particular area and arrest people, that becomes the area they target."
"The most awful thing is that people are flagged and put on a list," Matthias says. "We are so concerned about these programs because they are very anonymous -- and if the computer says so the computer will be right."
It appears predictive policing is even on its way to Canada. "I think this is the beginning," says Andrew. "We've entered this futuristic world where police are collecting lots of information about us, where we're being put into systems, where are neighbourhoods are being targeted."
As for what Matthias hopes people will take away from his film? "Be aware of what's happening."