What if it's Actually the Other Way Around? What if it's the "Higher-Skill" Jobs That Are Getting Automated First?

The title of this item may seem counterintuitive--not only to those who adhere to the conventional wisdom, but those who actually know something of the history of technology. The Industrial Revolution, after all, appeared to be about economizing the use of "muscle" rather than "brain," in part through the mobilization of more brain workers to design and build and operate the machines that replaced the earlier workers, as with, for example, the carding, spinning, weaving machinery of the eighteenth century that revolutionized textile production. However, the appearance is deceptive. The machinery replaced not just the physical effort but the mental effort--the mental skills--the learning--that the human workers put into the carding, the spinning, the weaving.

The latter may well have been the more important, and the trend more evident in this age of computer intelligence. As analysts of automation have long observed, down to Carl Benedikt Frey and Michael Osborne, "perception and manipulation," "finger" and "manual" dexterity, the ability to work in "cramped" spaces and "awkward" positions, have been significant bottlenecks to automation--while we take utterly for granted computers' capacity to perform vast quantities of complex calculations with a speed and accuracy far, far beyond that of the most able human. The result has been that the armies of mathematicians who served as "human computers" (like those brought to public attention by the film Hidden Figures) have been easily and quietly replaced by electronic computers in contrast with, for example, the janitors that keep the offices clean, and their less appreciated skills.

So it goes with the more recent wave of automation. The imminence of the self-driving car was oversold in the mid-'10s, such that in 2023 few expect to see them anytime soon. But there is great excitement over the capacity of chatbots like GPT-3 to, among much else, write code, to the point that there is much argument over whether coders will not become obsolete in the manner of human computers (while still more advanced and capable versions of the bot are expected before even the end of this year). If true this will mean that, in line with the bottlenecks previously discussed, artificial intelligence will have "learned to code" before it has "learned to drive"--fitting the aforementioned pattern all too well.

Granted, in all of these cases this elimination of "skilled" workers only went so far. If human computers were replaced, and coders might be replaced, this depended on persons with other skills conventionally regarded as higher still--the computer scientists who created the electronic computers (while in the latter case the rocket scientists the human computers supported remain very much with us). Likewise it is one thing to replace coders, another to replace "higher level" software engineers.

Yet even allowing for all of this the reality still complicates the conveniently hierarchical view so many take of these matters, enough that some rethinking of some widespread assumptions seems warranted. One is just what really should be thought of as constituting "high-level" skills--and the way in which our ideas about the intelligence required for particular forms of work reflects social prejudices (for instance, the tendency to denigrate those who have to move and "use their hands" as against persons in more stationary and less manual jobs). Another is that matter of which jobs will continue to be done longest by humans in an automating world--with the answer the exact opposite of the clichès. "Learn to code," the stereotypical, callous elitist sneers at truck drivers fearful they will be put out of a job--but it may well be that the pool of coding jobs will dry up before the pool of truck driving jobs does. Indeed, those really pessimistic about automated driving may expect that not just the coders but the software engineers will increasingly be put out of a job by improvements in AI even as humans go on driving those trucks. And of course, anyone who takes all this at all seriously should think long and hard about what all this means for glib talk about sending more people to college and more STEM, STEM, STEM! as the pseudo-thinking person's answer to all the woes of the work force.