External signal·Fortune·May 21, 2026·Tristan Bove·6 min read
A Nobel economist figured out 60 years ago that people learn best on the job. The Atlanta Fed says AI is making that almost impossible
“The tasks that fill entry-level positions are not merely low-value work—they are the curriculum through which workers accumulate the human capital that makes them productive later in their careers.”
Summary
A new Federal Reserve Bank of Atlanta working paper revives Nobel economist Kenneth Arrow's 1962 'learning-by-doing' theory to argue that automating entry-level jobs will eventually harm the firms doing the automating. The researchers contend entry-level tasks are not low-value busywork but the 'curriculum' through which workers build the human capital needed for senior roles. Because Arrow showed learning and productivity spill across the wider economy, one firm's automation choices ripple through its whole industry. With young degree-holders now facing unemployment at or above the overall rate, the paper warns that gutting the entry-level pipeline trades short-term payroll savings for a future shortage of capable senior workers. The authors propose taxing automation-derived profits while subsidizing firms that expand entry-level work.
Predictions for the future of work
The Atlanta Fed researchers predict that continued automation of junior office work will, over the long run, erode firms' pipelines of experienced senior staff and produce weaker managers and slower innovation, with the welfare cost falling 'almost entirely on workers.' They forecast that without policy intervention—an automation-profits tax plus subsidies for expanding entry-level tasks—companies will keep choosing short-term cost savings, pushing young graduates into unemployment or underemployment and depriving them of career-foundational learning.
Originally published by Fortune · May 21, 2026
Read the original at Fortune