The economists who told you not to worry about AI and jobs might be using the wrong playbook.
The Summary
- Economist Alex Imas argues AI may disrupt labor markets in ways fundamentally different from past technologies like the steam engine
- The standard economic model assumes tech displaces some jobs but creates new ones through productivity gains and emerging demand, but this framework may not apply to AI
- The question isn't whether there will be pain, but whether the historical pattern of labor market rebalancing still holds
The Signal
Every industrial revolution has followed the same script. Machines replace workers. Workers panic. New jobs emerge that nobody predicted. Equilibrium returns, just at a higher productivity level. Economists have watched this loop run for 200 years, so when people freak out about AI, the institutional response is: relax, we've seen this movie before.
Alex Imas is challenging that institutional memory. The Odd Lots conversation surfaces a critical question: what if AI breaks the pattern because it's not just automating tasks, it's automating the thing humans had left, the cognitive work we thought was our moat?
"Could AI be disruptive in a manner that, say, the steam engine was not?"
The steam engine replaced muscle. The computer replaced calculation. But AI replaces judgment, creativity, pattern recognition, the exact capabilities that let humans climb the skill ladder after previous waves of automation. When the steam engine took your factory job, you could learn bookkeeping. When software took bookkeeping, you could do customer success or design work. What do you learn when AI does all of that?
The standard economic model depends on a few assumptions:
- Displaced workers can retrain for new roles
- New sources of demand create jobs we can't imagine yet
- Productivity gains get distributed as economic growth that needs human labor to capture
That third assumption is where the model might crack. If agents can capture most of the productivity gains themselves, building more agents, serving more customers, writing more code, the flywheel spins without needing to hire humans at the same rate it historically did. The pain isn't temporary displacement. It's structural.
This matters because policy, investment, and corporate planning are all still running on the old script. Companies are investing in AI assuming they'll redeploy workers to higher-value tasks. Governments are assuming tax revenue will keep pace because new jobs will emerge. Workers are assuming the skills gap is temporary, solvable with a bootcamp or certification.
The Implication
If Imas is right, we're in the early chapters of a labor market transformation that doesn't have a historical parallel. The question for workers isn't "what should I retrain for" but "what kind of work has durable value when agents can do most cognitive tasks cheaper and faster." For companies, it's not just automation ROI, it's whether your business model still works when labor costs approach zero but consumer purchasing power doesn't grow at the same rate.
Watch what happens to white-collar job creation over the next 18 months. If new roles aren't emerging at the rate displaced ones disappear, the economists might actually be getting this one wrong.