Richard Mitchell, who once led the control design for a US nuclear plant, deliberately kept manual steps inside a fully automated system. A Harvard study of 65 million workers at 280,000 firms shows why that instinct matters again: entry level jobs fell about 9 percent after genAI adoption.
Richard Mitchell, a systems engineer and founder of AuraSpark Technologies, led the controls design for a first of its kind digital control system at a US nuclear plant just over a decade ago. The system was built to run itself much like a modern airliner, yet the team deliberately left manual steps inside sequences the automation could handle alone, so operators would not become mere observers. The plant was never built for political and economic reasons, but the design logic remains relevant: a Harvard working paper covering roughly 65 million workers at more than 280,000 US firms found that junior employment fell about 9 percent within six quarters after companies adopted generative AI, while senior employment kept growing.
A Stanford analysis of ADP payroll records confirms the pattern: the youngest workers in the most AI exposed occupations lost ground after late 2022 while more experienced colleagues held steady. The losses concentrate where AI automates tasks rather than merely assisting with them. For companies this means that cutting entry level roles because tools now handle routine work risks a gap in the next generation of leaders, since hands on experience cannot be built on a drawing board.
Causality remains contested. Researchers at the New York Fed attribute much of the rise in unemployment among young graduates not to AI but to remote work, arguing firms are more reluctant to train new hires from a distance. What stays open is whether companies are willing to build in deliberately inefficient training steps, the way Mitchell's team did for the plant project, or whether short term cost savings will keep winning out.
What this means for decision-makers
- Check whether junior staff still get hands on operating practice rather than just supervising automation.
- Set fixed manual training steps inside automated workflows that force regular hands on intervention.
- Compare junior and senior hiring numbers before and after AI adoption to catch displacement effects early.
This story was produced automatically from the source named above and checked by software before publication. The image is symbolic and shows neither the event nor a real person. How this paper is made