Machines That Know Us?

As creatures of habit, we humans tend to be very predictable.  For those of us who are married, we quickly learn behaviors and habits of our spouse so that we usually know how they might react to various situations, or whether they would choose one type of movie over another.  Familiarity aids us in developing the "sense" of knowing the other.

The ability to predict from observed behavior is something that is likely written into our genetic makeup.  It would logically be part of the set of survival tools that allowed us to become the dominant species on the planet.  We learned many, many generations ago that when certain things were observed -- e.g. a lion -- we needed to respond in a certain way -- run!  Developing the cognitive ability to recognize danger, or safety, or situations calling for decisive action were necessary for coping with the world around us.

Animals also appear to behave along the same lines.  People who are highly observant of how animals respond to stimuli are often extremely good trainers and achieve a level of effectiveness that appears almost empathic.  It is simply that they are observant of the cause and response mechanisms and able to employ them to affect behavior.

What sent me down this pathway today is an observation on Facebook by someone I grew up with that their automobile seemed to know their habits to the point it was spooky.  They got in the car to go to church at the same time they normally do on Sunday and their navigation system informed them, without any input, that it would take them 19 minutes to their destination of XYZ.  The immediate reaction was, "I think our devices know way too much about us!"

Pattern Recognition is a field of study that has been around for a long time.  In recent years computing ability has made it more accurate and more pervasive.  The field of statistics developed as a rudimentary way to analyze behavior and through mathematical modeling, predict likely outcomes based on past behavior.  Today, such modeling has been refined to the point that huge reams of data can be parsed into subsets which allow a tremendous amount of accuracy in predicting behavior.

In the instance of the example above, it is clear that the machine (their GPS) has the ability to recognize patterns at some level.  If every Sunday afternoon at 5:40 they start their car to go to church and approximately 19 minutes later they arrive there, a clear pattern begins to emerge.  If next Sunday they get in that car and start it at approximately 5:40, where are they likely going?  You get the picture.

Computers have given us the power to model behavior in ways never before imagined.  It will be interesting to see how more and more machines mimic human behavior.  It isn't so much that the machines are exhibiting intelligence as they are using predictive algorithms to solve problems almost instantaneously that analyze the situation and compare it to a database of behavior-response scenarios that allow the prediction of likely outcome based on a specific action.

Alan Turing once wrote on the subject.  He suggested that we shouldn't ask the question, "Can machines think?" rather, we should ask, "Can machines do what we (as thinking entities) can do?"  It seems the answer is yes.