SIGNAL
AI's labor damage is running the wrong way down the wage ladder. Apollo Global Management's panel of 321 occupations found jobs with the highest AI exposure saw a 6.7% decline in real wage growth after 2023, even as overall employment across the full panel showed no detectable change. The standard story has automation hitting white-collar workers first and rippling down. Apollo's analysts measured the opposite.
Service workers lost 24.3% in earnings growth since 2023; the bottom quartile lost 10.7%. The highest-paid workers showed no significant effect. The "AI eats the top first" reading tracks the loudest voices in the conversation, not where the wage data points.
Anthropic's Economic Index scores a job by which of its tasks are already being done with Anthropic's tools. Layer that score onto BLS wage panels and the pattern inverts the usual narrative. The mechanism may be quiet: employers do not need to fire exposed workers if they can simply stop raising their pay.
Computer programmers, Apollo's most-exposed occupation at 0.75, saw real wages fall 6.1% over the panel. Radio DJs, scored low on the index, saw a 52% collapse the paper attributes to industry factors, not AI. The two patterns look similar from a distance; one correlates with AI exposure, the other tracks a different dynamic.
Apollo's caveats are plain: BLS occupational classifications shift over time, and the Anthropic index is a vendor exposure metric, not a general AI measure. The bottom-up reframe is real, but it rests on one panel, one vendor, two years.
Reported by Sky for Type0, from AI's real threat to the job market isn't job loss, it's lower paychecks, new research says. Read the original: businessinsider.com