After free trade and automation hit certain workers hardest, will AI's economic hit land more evenly — or just shift who absorbs it?

This explores what history tells us about how workers fared after earlier economic shocks like free trade and automation, and whether those lessons apply to AI. A direct note first: the collection doesn't contain studies of free trade itself, such as the China trade shock or NAFTA. What it does have is a set of economists comparing AI with earlier waves of automation. That comparison turns out to be the more revealing one.

The clearest historical anchor is Acemoglu's modeling Will AI increase inequality less than past automation?. Earlier automation hit specific groups hard, especially workers without college degrees doing routine tasks. AI's wage effects look more evenly spread across demographic groups. The catch is that even when losses are shared more evenly, income from owning capital still pulls ahead of income from working. Anthropic's scenario modeling reaches a similar result Does AI growth inevitably shift wealth away from workers?. Average wages can rise while workers' share of the economy shrinks, and knowledge workers' pay stalls. The economy grows, but the gains don't automatically reach the people whose work changed.

Why didn't workers simply bounce back in past shifts? The corpus points to bargaining power rather than the technology itself. Brynjolfsson argues that technology built to replace workers, rather than make them more capable, shifts leverage to whoever owns the tools. That can lock in a stable trap where workers can't push back Does automating labor concentrate power more than augmenting it?. A related idea goes further and treats human labor as a hidden safeguard Does incremental AI replacement erode human influence over society?. Institutions stay responsive to people partly because they depend on people to run them. Remove that dependence and you lose more than wages.

The more hopeful lesson from past transitions is that workers often move into the parts of a job that weren't automated. Firm-level data from 2010 to 2023 bears this out Does concentrated AI exposure enable workers to adapt and reallocate?. When AI touches only a few tasks in a job, workers shift to the rest, and net job losses stay small. Adjustment also isn't uniform. Some firms replace contract workers with AI much faster and more cheaply than others Do firms substitute labor for AI at different rates?. In female-dominated occupations, exposure reaches lower-paid workers who have fewer resources to adapt Does AI exposure hit low-wage workers harder in some fields?.

Here's the part you might not expect. Past shocks often showed up first as fewer jobs being created, not as mass layoffs, and AI seems to follow that pattern. Payroll data show no economy-wide job losses so far. Young workers in AI-exposed fields, though, are being hired at 19% lower rates, while experienced workers see no such gap Is generative AI displacing workers at economy-wide scale?. Workers seem to sense it: in a survey of 81,000 users, early-career people worried most about losing their jobs Does AI productivity gain always ease job displacement fears?. If you want the historical comparison with free trade specifically, you'll need sources outside this collection. If you want to know which of those lessons might carry over to AI, the notes above are where to start.