Is AI exposure reducing employment for the youngest workers?
Core argument: Employment of early career workers aged 22–24 in the most AI-exposed quintile of industry-state employers fell 15.2% from 2022q4 to 2025q2, while all other exposure quintiles were flat or posted modest gains.
Figure 3 provides a comparison of total employment among early career workers [those aged 22-24] across quintiles of industry-state AI exposure. It shows that early career employment in the most AI-exposed quintile of employers declined by 15.2% from the reference period of 2022q4 to 2025q2. All other quintiles either had flat employment or exhibited modest gains over the same period. This result is considerably larger than the 6% decline in employment among the most exposed workers as reported by Brynjolfsson, Chandar and Chen (2025). However, since I also find weaker employment growth among less-exposed quintiles, the unadjusted gap in employment growth is similar in magnitude. For example, although the result of Figure 3 is not regression adjusted, it is highly consistent with the 16% regression-adjusted relative decline in employment that Brynjolfsson et al. observe among the most exposed occupations in ages 22-24. Figure 4 [] compares most exposed quintile employment changes for early career workers against those of workers in four age bins of older workers. This figure demonstrates that the decline in employment of early career workers in AI-exposed industries is not generally observed among older workers. The 25-34 age group saw a substantial decline in employment over this period, but in percentage terms it was less than one third as large as that observed among ages 22-24.




