New Work In The Age Of Ai Doesn’t Need A Government Planner

The AI jobpocalypse crowd has a fresh challenge to its view. “What Makes New Work Different from More Work?,” a new NBER working paper by economist David Autor and co-authors, shows that the American economy keeps spinning off new job specialties that pay a premium for scarce, newly demanded expertise: 

We argue that new work is central to maintaining and augmenting the value of human expertise in the face of automation because, unlike simply more (preexisting) work, new work demands novel expertise that commands a scarcity premium. … These findings suggest that new work serves as a countervailing force to automation-driven job displacement not merely by creating additional employment, but also by generating new domains of human expertise that command market premiums. 

That’s a useful corrective to automation doomsayers, who often fall victim to the “lump of labor fallacy”—the idea that there’s a fixed amount of work to be done, and every task claimed by a machine means less for humans to do. In fact, the job market is always evolving. Jobs appear and disappear. As the paper points out: 

Between 1920 and 1940, for example, the monopoly operator of the U.S. telephone network, AT&T, implemented a staggered adoption of mechanical telephone switching technology across U.S. cities. As cities made the technological transition, their local telephone operator employment fell by 50 to 80% within a decade.

Let’s focus on new work, which is the point of the paper. Using newly available Census data from 1940 to 2023, the researchers find one in five American workers today holds a job that didn’t exist before 1970. What’s more, “new work commands a wage premium even within detailed occupations and industries, controlling for demographics and human capital.” That premium is largest for technology-linked roles, by the way, and it fades as skills spread—basically what you’d expect if scarce expertise drives the returns.

One conclusion: It matters whether AI not only automates tasks but also augments workers in ways that generate new tasks and new demand for human skill. (The pessimists typically miss that second part). And if that’s true, is there maybe a role for government in directing AI so it augments as well as automates?

The trouble is that technology almost never works that neatly. Its effects on jobs are messy, unpredictable, and shaped by millions of decisions from businesses and entrepreneurs—not by a multipoint plan fashioned by government. Trying to design technology around specific job goals risks locking in today’s assumptions instead of leaving room for entirely new kinds of work, the kind that often appears only after a technology spreads. 

As economist Joshua Gans has noted, the concept of “pro-worker” technology is itself unstable. Skills are valuable partly because not everyone has them. Spreading them widely may help more workers overall while reducing the pay advantage of those who once had rare expertise.

Anyway, Autor is offering a historical argument. No human or machine can prove AI won’t break the pattern he’s identified in the coming years and decades. Predictions are hard, especially about the future. But let’s be clear: Those pushing for industrial policy and dramatic new government interventions are making a big forecasting leap—that the pattern will surely break, and that government can fix it from the top down.

So what should policymakers do? Not try to slow automation, but speed adaptation. As Bloomberg columnist Clive Crook neatly argues, taxing AI or robots risks choking off the productivity growth that rising incomes depend on. The smarter path is to promote augmentation: a) invest in basic R&D, b) reform education and training so workers can use AI effectively, and c) make it easier for workers to move into new kinds of jobs. Success in that final point means tackling hidden barriers like excessive occupational licensing, which blocks capable workers from new roles. The goal isn’t fewer machines, but more—and better—jobs where humans use AI to do new, higher-value work.

If AI really is a powerful general-purpose technology, it will boost worker productivity. Higher wages should eventually follow—the most pro-worker outcome of all.

The post New Work in the Age of AI Doesn’t Need a Government Planner appeared first on American Enterprise Institute - AEI.