The most disruptive technology in human history was almost certainly the wheel. That economic shock, and all the others that followed, give us useful insight into labor market effects of artificial intelligence.

The invention of the wheel cut transportation costs by 80% or 90%, dramatically reducing demand for workers who carried goods across and between towns. We’ve had other technology shocks—the use of fossil fuels, steam and then electric power, the internal combustion engine and computers.

All these technologies replaced tasks that were part of jobs. The wheel replaced a strong back, the steam loom replaced strong legs. The use of fossil fuels replaced the cutting, splitting and drying of lumber, and electricity replaced the use of steam looms.

For us, the computer has been the most disruptive technology. It radically changed the types of work that almost everybody performs. It also changed our ways of communicating, our amusements, our safety and health. It brought us the internet, social media and now AI.

AI has been around in some form since the 1950s. I first heard about it in 1992, when a colleague of mine, then an infantry captain, was sent to obtain a master’s degree in AI. By 1997, I was learning the use of rudimentary AI in economic modeling.

The new, commercial applications of AI are much more advanced—and interesting—than the early AI algorithms of the 1990s. The large language models (LLM) are superb for writing reports, school papers and summaries of some topics. Generative AI can construct pictures and movies that are almost indistinguishable from the work of actual humans.

The potential applications of these new technologies are boundless, to the extent that any one person could predict. I see all types of uses in economics and warfare, the two fields I’ve been trained and educated in. There are also limitations.

I’ve asked commercial versions of LLMs to provide novel testable hypotheses in economics—the lifeblood of economic analysis. The LLMs are good at naming data sources and, with enough prompts, can even construct the mathematical model to support a hypothesis. But none of the hypotheses were really any more than most middle school kids could have derived.

The generative AI models are equally poor right now, delivering pictures of people with seven fingers or grilling burgers with lettuce, tomatoes and buns. They’ll get better, of course, but what is AI likely to do to the demand for labor?

I think the easy answer is that it will increase the demand for labor, in much the same way as the wheel, the steam loom, the automobile and the computer. That is, in a very nuanced way.

Technology doesn’t replace jobs; it replaces tasks. Almost always, the tasks replaced are the most mundane, routine and trainable ones. In so doing, the technology makes the uniquely human part of the job more valuable.

The best long-form description of this comes in an accessible paper by David Autor (https://www.nber.org/system/files/working_papers/w20485/w20485.pdf) who described Polanyi’s Paradox, that “we can know more [about our jobs] than we can tell.” The point of Polyani, which Autor fleshed out in superb contemporary detail, is that the unseen part of technology is how humans adapt it to complement their innate skills.

Since the end of World War II, technology has replaced more than 80% of the work done by the average American. Throughout the longest and most impactful technology shock, the U.S. boosted wages, production and employment.

AI may be different than any technology before it, but the adaptation was not technological—it was human. We humans are much as we’ve always been, and the economic incentive to match complementary human and technology skills remains robust.

The most likely outcome of AI adoption will be positive, like all the other technology adoptions before it. But that doesn’t mean there won’t be challenges.

The most dramatically unpleasant periods of technology adoption occurred in the places, and among the people, that could not adapt. James Whitcomb Riley’s The Raggedy Man (https://www.poetryfoundation.org/poems/44955/the-raggedy-man-56d2243f915f3) of 1888 described a type of itinerant worker that existed until at least the 1960s in U.S. agriculture.

The Raggedy Man is gone now, because the skills he brought to a farm are no longer sufficient to earn him three meals and a simple room. Even then, one is tempted by this poem to conclude that he was employed for reasons beyond labor productivity.

Technology eliminates the less skilled tasks a worker does, pushing them to more skilled—and more uniquely human—tasks.

AI is likely to impact skills held by more educated workers than the robotics of the 1980s and later, or the digitization of the 2000s. AI will write simple research summaries, press releases and perform straightforward design work. This will lead to increased demand for more detailed and complex research summaries, more insightful press releases and more innovative designs than AI can produce.

AI will also open demand for employment totally divorced from the direct complementarity to technology. As easily replicable human skills become inexpensive, the relative value of scarcer, purely human skills will rise.

What does AI portend for education and regions?

The one common thread of all previous technologies is that they complemented human-specific intellectual and social skills. So, job losses were clustered among those who were armed with skills that were more readily replaced.

Thus, AI is likely to boost demand for workers with a lengthier, broader and more complex education. That education accesses more latent human skills. This used to be called a liberal education, but a better moniker is a classical education.

Of course, such an education is not trendy today, in part because it is costly. It is much cheaper and faster to prepare for the last technological shock than the next one.

---
Note:  The views expressed here are solely those of the author, and do not represent those of funders, associations, any entity of Ball State University, or its governing body.