The Star’s Erika Bolstad reports on the early use of drones in South Africa to combat Rhino horn poaching, which is decimating Rhino populations at alarming rates. The computational work behind this, spearheaded by Tom Snitch at the University of Maryland, is promising because he will fuse together all manner of contextual data, from weather patterns and moon phases to historical poaching statistics, to have the drones create the best possible estimate of likely poaching locations and times. This is a little tiny nugget of crime future-casting, and the hope is that it can lead to positioning the drones at the right places and the right times to intercept poaching operations before Rhinos die, rather than just punishing criminals after the damage is done.
This use of drone and intelligent computational models makes good sense for vast geographically dispersed problems like poaching; but the downstream consequences apply to urban areas as well. Take the article and replace “Rhinos” by “gang warfare” and imagine the same “AI” applied to human-on-human violence. What if drones, powered by computer muscle, would fly to the likeliest locations where crimes are about to happen within our city boundaries? Accountability takes a strange step sideways when drones circle your house on friday night, you complain to the local constable, and he explains that they’ve decided crime is likely here thank to sophisticated AI algorithms. Sorry, but it’s not a human decision- the computers are right often enough we trust them on this. It might put a bit of a damper on that backyard barbecue party you are having for your neighborhood that evening.
Robotics and data processing promise more decision-making in the hands of computers than ever before. Rhino poaching is an excellent example of a positive application, and yet every such application will also lead to muddier waters as well- that’s in the nature of the robot futures we are approaching.