US Bureau of Labor Statistics data indicate unemployment rates for occupations with high AI exposure remain below those for less-exposed roles, with no measurable occupational shifts toward manual labor.

MIT Technology Review analysis of BLS figures through 2025 finds no large-scale displacement, citing Erika McEntarfer that only one in five firms report AI use in any function per US Census Bureau data; this aligns with historical lags documented in Autor, Levy and Murnane (2003) where computerization effects required decades to alter task structures across industries.

Recent BLS occupational employment statistics through Q4 2025 show software developer and analyst roles holding steady hiring rates absent mass reallocation, contradicting claims of rapid white-collar erosion while echoing overstated automation forecasts from the 2010s that projected manufacturing job losses far exceeding realized outcomes per Acemoglu and Restrepo (2020).

Younger worker unemployment at 5.6% for recent graduates correlates more with post-pandemic hiring slowdowns than AI per BLS JOLTS series, underscoring the need for firm-level adoption metrics before labor market inference.