Judging a computer algorithm

Does your GPS have a name?

Years ago when I bought the first GPS unit for our family we gave it a name within a week. There was a human voice. That voice gave us instructions and told us when we made mistakes and needed to correct our route. It was like having a live person in the car with us. The kids liked the naming process because we tried to match a name with the voice that came from the device.nuvi-265wt

But something else happened.

We started to rate the instructions we received from the device. We would get a little excited when we knew multiple ways to navigate a route and preferred another option from what the GPS determined to be the best route. We would talk to the device as if it could understand us and tell it that we knew a better way. We would stay on the larger interstate when the GPS wanted us to exit and drive on smaller two-lane roads. .

But why?

It’s just an algorithm in a computer program. It’s performing math calculations to determine the shortest route between two distances. This particular version of the GPS did not have current information related to traffic patterns that factored into the algorithm. Yet we wanted to treat the unit like a person. We judged the instructions and the outcome and expected the GPS to know about new roads and road detours.

Do we judge algorithm mistakes differently than we judge human mistakes?

There was a recent story on NPR from Shankar Vedantam in which he discusses research from the University of Pennsylvania about human’s use of algorithms. The research found that humans will typically stop using a computer algorithm after they experienced a mistake even though the computer algorithm was less likely to make a mistake than a human counterpart.

Vedantam goes on to discuss that algorithms are getting smarter and more complex. They learn from mistakes. They have the ability to make decisions based on a variety of inputs. So it’s illogical to judge the algorithm more harshly than a human but yet it feels so natural to many people.

Do we feel threatened by algorithms? Perhaps movies like The Matrix or Ex Machina make us more aware of just what a truly learning algorithm could be like. Maybe that’s what influences us to judge simple computer programs more harshly.

I will say this.

When the GPS takes us on a tangent, I’m always the one saying “Trust her. She’s always gets us there.”  I stand by her instructions even though she’s taken me to a few dead-end roads.

But I’ve never trusted Siri. :-O

Onward and upward!