Bloomberg news reports that some hi-tech companies have begun to develop multimedia programs to assess a person’s suitability for a particular job, or jobs. Presumably, these programs will be much better than human recruiters, who rely on first impressions, useless tools, and too many popular but invalid assumptions about what makes a good employee. The new, computerized models will analyze every millisecond of an interaction with a potential employee, ferreting out skills and tendencies that can then be optimally matched to available jobs that an applicant would be best at.
There is something to be said for this. First impressions count, but they also tend to be off the mark, since they rely too much on appearances. Anyone can be impressive for an hour or so, especially folks who specialize in fooling folks. Popular tools like the Myers-Briggs test have so scientific validity. And assumptions like, for instance, a correlation between employee unreliability and a previous history of job hopping, are not supported by data.
On the other hand…
In my view, computer programs and models are very useful in many applications, but they tend to fail when they must predict not just human behavior but any phenomenon that contains an unknown number of unpredictable variables. The running battle over global warming predictions offered by models is just one notorious example.
Famously, people are predictable until they are not. That’s the problem with that pesky business of human free will, motivation, desire, and so forth. I suspect that the program, no matter how sophisticated, will not detect the fact that, for instance, I am just interviewing for a dream job of a lifetime and that my job-hopping days are over if I get it; and even if the machine says that I am not very good at it now, my motivation is such that I will soon master it.
Don’t get me wrong. I am in awe of Mr. Data, and on most things, I would follow his advice without reservation. But when it comes to ever so slightly lesser machines, well, not so much. It may turn out that the assumptions on which these new systems are constructed are just as flawed as those of the old-style interviewers and headhunters.