Is This The Age of Specialization?
I usually write tech related articles. This one is slightly different in the sense it uses a few examples and analogies from tech to illustrate a few points but is not exclusive only to the tech space and can be said to be ubiquitous across other disciplines as well.
The generalist in any field is typically someone with a wide array of knowledge, someone with the high-level knowledge to either troubleshoot or diagnose a problem. In medicine, you get the General Practitioner (GP) whereas in tech you get the Tier I (or alternatively L1) support responsible for basic customer support issues. With the advent of Robotic Process Automation (RPA) and AI the impact of disruption on traditional job roles is going to be significant to say the least. The way I see it, it is most likely the generalist in any field who is going to find his or her job at risk. The reason for this being it is usually the generalist eliminates and narrows down possibilities algorithmically, and as AI brings in the capabilities of consolidating and deriving answers across data and RPA brings in the capability for automation, he or she is in the greatest danger of losing their job.
Now consider the specialist. He or she is an individual who knows “more and more” of “less and less”. (Polymaths excluded, of course.) IMHO specialist may indeed start out as generalists but over the years as they focus on a niche area, their knowledge on the other areas tend to be restricted to the peripheries due to perhaps getting atrophied. However, in their niche area they are the experts, the ones who define the industry best practices etc .,. Usually when we develop predictive models in AI, say for example using an MRI to identify an aneurysm (an example off the top of my head), we need to use a gold standard or baseline developed on a labeled training data set. In the example given, this would be a set of patient MRIs annotated by a neurologist or group of neurologists as either having an aneurysm or not. For the predictive model to be successful it would have to be on average at least as accurate as predicting an aneurysm on a new set of MRIs compared to the predictions made by a group of neurologists. Still, that’s not the end of the problem as I see it. As an expert in his or her field a specialist owes a higher-level of “duty of care” towards their customers than a generalist (at least IMHO) when they are consulted. They take on a higher level of responsibility and are held liable as experts to justify their comments/diagnoses. As such in the example above say that a patient with an aneurysm dies because of not identifying the aneurysm on the MRI through the predictive model (an oversimplification of events but bear with me), who is going to be held responsible for the death?
In a somewhat pedantic way what I have tried to illustrate above is that if the specialist owes a greater “duty of care” towards his customers/patients than a generalist, the chances of his or her job being automated in the long run are not significant. I also feel that if the rate of automation were to increase faster than the rate of which specialists are churned out that the negative social externalities which might occur could outweigh the benefits of automation. What do you think? I would love to hear your comments!