It turns out the richest countries are actually the least likely to use AI to replace human workers compared to poorer nations.
April 13, 2026
Original Paper
The Simple Macroeconomics of AI: An Empirical Test Using the Anthropic Economic Index
SSRN · 6274118
The Takeaway
While we fear a total job apocalypse, data shows that high-income countries tend to use AI to augment and enhance human labor rather than swap it out. This suggests the link between automation and job loss isn't a straight line, and prosperity might actually protect workers.
From the abstract
This paper provides one of the first large-scale empirical tests of Acemoglu's (2024) "Simple Macroeconomics of AI" framework using revealed-preference data from the Anthropic Economic Index v4 (November 2025). Covering 2,134 O*NET tasks across 118 countries, we test four hypotheses derived from Acemoglu's task-based model: (H1) the relationship between national income and automation intensity, (H2) whether complex tasks favor augmentation over automation, (H3) whether AI enables identification