The Hidden Cost of AI: How Automating Junior Work May Erode Expertise

Summary

This opinion essay argues that agentic AI is removing the entry-level work that traditionally trained white-collar professionals. Drawing on the Air France 447 crash and Lisanne Bainbridge’s “Ironies of Automation,” it says automation can leave humans responsible for exceptional or emergency decisions while depriving them of the routine practice needed to handle those situations. The author describes junior work as a hidden cross-subsidy: firms received useful output while also producing their next generation of experienced employees. AI separates those products by producing the output more cheaply, without automatically replacing the training. The essay uses Polanyi’s tacit knowledge, the Dreyfus model of skill acquisition, Bainbridge’s automation paradox, and Becker’s account of underinvestment in transferable skills to argue that expertise cannot simply be downloaded and that individual firms have weak incentives to fund it. It cites data showing no overall collapse in white-collar employment, but sharper declines in early-career hiring at AI-adopting firms and in AI-exposed occupations, along with employers asking for more prior experience. The author acknowledges that AI could compress the learning ladder if used as a supervised source of deliberate practice, similar to a flight simulator. The proposed response is to design and fund apprenticeships deliberately, move junior workers into judgment-focused roles, measure talent pipelines, assign accountable owners, and develop shared training systems across industries.