No, Computers Aren't "Smart" At All

Computers are fascinating and, at least to those of us born during a time when “laptop” meant a mobile desk intended for easing the process by which microwaved meals could be consumed in front of a television, intimidating.  The advent of intelligent software stirs up irrational fear instilled in us from movies such as The Matrix, the Terminator series, and I, Robot.  Due to my parents’ dubious judgement as to which VHS tapes I should have been allowed to dig out of the family collection and watch by myself, hearing the term artificial intelligence (AI) automatically makes me envision the   software (Siri, Alexis) brings to mind HAL 9000, the evil supercomputer from 2001: A Space Odyssey nonchalantly proclaiming “I’m sorry Dave, I’m afraid I can’t do that.”  On a more rational level, concerns regarding recording and tracking of our conversations by the government and private companies for public safety, advertising and perhaps more sinister motives have increased since we can now purchase “spies” in the form of computerized personal assistants from Apple and Amazon.  An article published by David Auerbach in the Winter 2013 issue of N+1 Magazine titled “The Stupidity of Computers” substantially allays these apprehensions by assuring readers that computers—while perhaps our betters when it comes to mathematical processing and data storage and retrieval—are considerably lacking in real, applicable “street smart” intelligence.

Auerbach provides an overview of the history (and predicable long-term future) of computers and their evolvement into artificial intelligence.  He proclaims that computers are “stupid” simply because they are incapable of navigating the nuances and complexities of human communication, which is more than face value of language. Human-to-human communication is a synergistic compound of language, tone, vernacular, culture, and situational context.  In order to “understand” human language, computers must systematically categorize language in terms of most-common (or common to the geographical region) usage.  Computers make a superhuman-strength informed educated guess as to what humans intend to mean in the use of language, but it is a guess all the same.  Therefore, computers often “guess” incorrectly due to variances in language structure, intended meaning, and—most typically—context.  As Auerbach argues, computers must constantly rely upon humans to form their “decisions,” as evidenced by the mistakes they often make. 

After a series of feedback prompts after issuing the command to “find category,” frequent virtual assistant users are quick to figure out that being specific helps Siri and Alexa help them.  Too vague of keywords will result in the virtual assistant searching data sources for too broad of a category.  A recent personal experience illustrates how computers are “dumb” in this regard.  A neighbor informed me that the dog park in our area had been given a recent overall with many improvements.  Having only visited once (over a year before) and did not commit the location of the park to memory due to being unimpressed with the condition at the time, I enlisted the aid of my new Garmin GPS navigation unit to help guide my eager younger dog there.  After asking numerous times to direct us to the dog park, the polite female voice emanating from the Garmin could only alternatively answer with “here are the pet stores in your area” and “directions to Ruffner Mountain Nature Preserve.”  This was really frustrating, as I had no need to go to Petco and no desire to go hiking at the local nature preserve.  Now, most humans familiar with the area could direct me to where I wanted to go, but the unhelpful complex unit of artificial intelligence did not possess the wherewithal to decipher a simple term comprised of two very simple English words, “dog” and “park.”

 Auerbach’s takeaway message is fairly simple: people need to exercise the extent of their dependence on computers with caution.  Relying too heavily upon computers for assisting us with our daily interpersonal interactions is a “human error.”  After all, who goes to a stupid source of information for daily advice?  And as Auerbach puts it, although they can be used for nefarious purposes, computers are simply too dumb to take over the world on their own.