AHEADE OF THE CURVE: Modelling human behaviour: Insight won’t compute

AHEADE OF THE CURVE: Modelling human behaviour: Insight won’t compute

Since the rise of data processing in the ‘50s, efforts have been made to apply technol­ogical models to human thought and behaviour. Scientists have drawn direct associations between thought and calculation, human memory and storage (temporary and persistent), telecommunications, and interpersonal contact. What mysteries of humanity might we unravel if we treated the three volumes of Donald Knuth’s The Art Of Computer Programming as psychology textbooks?

This subject was brought to my mind by a paper on Latent Semantic Analysis (LSA). LSA swallows a huge blob of text then performs comparative analysis of other documents. The authors are careful not to make outlandish claims about their research. They do claim that it approximates our learning process well enough to match the accuracy of humans at certain types of evaluative work. It can grade essays as well as a teacher, and it is used to screen patients for depression.

A computer can likely replace an inept teacher or a first-year medical resident, working solely from statistically derived models of mental illness. But a computer cannot, and never will, have a feel for the work. Insight, passion, and empathy can’t be taught to people, so they certainly can’t be programmed into a computer.

Insight is not an analytical process made more efficient by repetition or by the acquisition of ever-larger pools of knowledge. It is the combined application of all resources, intellectual and sensory, to problem identification and resolution. It is the reason your first answers are usually correct.

It explains why putting more hours into a project, or thinking harder about it, rarely moves you closer to completion. To me, phrases such as “can’t see the forest for the trees” refer to the innate competition between computation and insight.

There are doctors who can look at you and immediately know you’re sick. Those inclined to believe in the unique properties of humanity would say the snap diagnosis is a knack. It’s a quality unavailable to those doing a job that pays best rather than one they were born to do.

In contrast, reason dictates that doctors who make accurate visual diagnoses are able to do so because they have memorised external signs of illness. In residency, these doctors saw enough examples of jaundice, sleep deprivation, and malnutrition to spot those conditions and others immediately.

If working by the book is the limit of a physician’s abilities, I submit that he or she belongs behind a microscope, not in a treatment room. If a psychologist evaluates and counsels patients by following a flowchart, that person should be running rats around a maze instead. If papers are graded by a machine, students will write to increase their chances of favourable semantic comparison. They’ll be conditioned to write to please the algorithm. I can’t make myself imagine what such writing would look like. If my kid’s English teacher is a kiosk, I’ll be home-schooling.

I’m confident that we’ll eventually have a vast computational grid that rivals the speed and capacity of the brain. Not in my lifetime, but someday. Then scientists will model emotional responses based on complex statistical matrices built from observed behaviour. I’m just as confident that the sun will burn out before anyone imparts insight to a computer.

Follow Us

Join the newsletter!

Error: Please check your email address.
Show Comments