Remote work or AI? A second cause of the shrinking junior hiring share

Peter John Lambert and Yannick Schindler, “The Broken Ladder: AI, Remote Work, and Early-Career Hiring,” CAGE Working Paper 808/2026; authors’ short 17 August essay. Career-path question. They examine 243 million new-hire records and 407 million job postings across the United States, UK, Canada and Australia, 2017–25, and classify firms as remote-work adopters using postings that offered remote/hybrid work in 2021–22. Junior share of hires fell substantially across the four countries; AI-exposed occupations also tended to be remotely work-feasible (exposure correlation reported as 0.77). In joint comparisons, their AI-exposure association attenuates after controlling for work-from-home exposure, while work-from-home associations persist; remote-adopting firms subsequently show a lower junior-hire share than matched nonadopters.

Causal mechanism offered: if early training relies on informal supervision and feedback, remote organization can make novice hires more costly; AI may simultaneously make some starter tasks less necessary. This is a credible competing explanation for exposure-only correlations, not proof that remote work rather than AI caused the full decline. Job-ad remote offers proxy actual use; firm preference for seniors, industry demand and pandemic-era recruiting can select remote adopters. Hires and postings are platform-visible rather than complete payroll records; junior share is not junior headcount and neither establishes a graduate’s later earnings. The authors explicitly allow a double effect and acknowledge AI effects persist in some payroll studies that control for remote-work intensity. An authorized future trial would compare matched teams given tested mentorship and AI adoption interventions on junior skills, output and retention.