Robot Employment Survival 101
Fear not the rise of the machines? That appears to be the advice given by MIT economist David H. Autor in a paper he recently presented at the Kansas City Fed’s symposium in Jackson Hole, Wyoming. Responding to a significant uptick in economists’ concern over the effects of automation on employment, including the “stunning” results of a poll suggesting “that a plurality of mainstream economists has accepted—at least tentatively—the proposition that a decade of technological advancement has made the median worker no better off, and possibly worse off,” Autor suggests that it will be a lot harder to automate us all away than many journalists and expert commentators have indicated.
Autor argues that “Polanyi’s paradox,” whereby “We can know more than we can tell…” saves us from the threat of total automation dislocation, because there will always be jobs that rely on a variety of particularly human skills and tasks, skills and tasks that we can’t entirely explain to ourselves, much less to a computer. As the philosopher Michael Polanyi himself put it, “The skill of a driver cannot be replaced by a thorough schooling in the theory of the motorcar; the knowledge I have of my own body differs altogether from the knowledge of its physiology.”
While jobs consisting almost entirely of routine tasks, i.e. those easily codified into rules that can then be automated, have been and will continue to be replaced by machine labor, then, there is according to Autor a natural buffer to keep many people employed (if not necessarily well paid). In fact, Autor sees a significant opportunity for computer-enhanced human labor, for “tasks that cannot be substituted by computerization are generally complemented by it.” The construction worker, in his example, has to manage too many variables in a fluid environment to be automated away. However, he can be given a backhoe to replace his shovel, enhancing the productivity of his labor while making backhoe-trained workers more valuable than the merely shovel-ready.
This is a blue-collar example of “skill-biased technical change,” more traditionally described by Autor’s fellow MIT professors (and techno-employment pessimists) Erik Brynjolfsson and Andrew McAfee:
Technologies like robotics, numerically controlled machines, computerized inventory control, and automatic transcription have been substituting for routine tasks, displacing those workers. Meanwhile other technologies like data visualization, analytics, high-speed communications, and rapid prototyping have augmented the contributions of more abstract and data-driven reasoning, increasing the value of those jobs.
Brynjolfsson and McAfee discuss in their book a polarization of the employment market, where high-skill abstract-task intensive jobs are increasingly well compensated, and well complemented by machine labor. Low-skill Polanyi paradox jobs, like janitorial work and home health care, are also insulated from being automated away, but as Autor describes it, they are too well-insulated to even benefit from automation complementing their labor. Because their jobs require only the minimal amount of human reasoning that any competent adult can provide, their wages are depressed by the large supply of interchangeable labor. Middle-skill jobs, however, are nearly wiped out in Brynjolfsson and McAfee’s analysis.
Here, too, Autor finds some reason for more optimism. He concludes that “employment polarization will not continue indefinitely,” for “While many middle-skill tasks are susceptible to automation, many middle-skill jobs demand a mixture of tasks from across the skill spectrum.” Moreover, “many of the tasks currently bundled into these jobs cannot readily be unbundled—with machines performing the middle-skill tasks and workers performing the residual—without a significant drop in quality.” Autor’s example here is the technical support call center where a human is retained as a social conveyance device for the troubleshooting heuristics of the computer system sitting in front of him. That may seem like efficient low-skill complementarity, but it in fact turns out to be very frustrating to discover that the technical support person has no knowledge, creativity, or initiative beyond what the computer tells them to read. Autor says “this is generally not a productive form of work organization because it fails to harness the complementaries between technical and interpersonal skills.”
Both Autor and Brynjolfsson and McAfee describe how systems are redesigned in order to take advantage of automation, however. Brynjolfsson and McAfee wrote that “a key aspect of SBTC was not just the skills of those working with computers, but more importantly the broader changes in work organization that were made possible by information technology.” They continued, “It was not so much that those directly working with computers had to be more skilled, but rather that whole production processes, and even industries, were reengineered to exploit powerful new information technologies.” Autor gives the example of Amazon, formerly reliant on low-skill runners to pick their products for shipping, dashing around the warehouse, bringing in Kiva Systems to design a more robot-friendly system where the shelves were programmed to come to the pickers, reducing the human job to only that task that could not be exported.
As the automated economy progresses, we would do well to remember that, while there is certainly a baked-in pattern and logic to computerized work, the programming away of jobs is performed by people. Programmers and management consultants decide how best to “reengineer” “whole production processes, and even industries” to take advantage of the capabilities of computers. That it so happens that jobs resembling those of the programmers and consultants, jobs high in abstraction, turn out to also be those best complemented, rather than replaced, by automation may be more than convenient.
The seemingly mundane routinized tasks of the Amazon picker, or the retail worker, or the technical support specialist may seem to be little more than drudgery and tasks to be automated. But, as Autor describes in several contexts, there are many ways in which those routine tasks can be embedded in more comprehensive work environments, and computerization can add value to the job as it stands. The trick is, the programmer and the consultant have to see the job in all its particularity before they could know how to complement it. They rarely do, and so those jobs are rarely complemented. They have to see how the support specialist uses his whole repertoire of human advantages before they replace him with a screen reader.