
Survey of Business Uncertainty
Federal Reserve survey weighting AI adoption by employment; produced 78% adoption rate for late 2025 — the highest federal measure.
Last refreshed: 16 April 2026 · Appears in 1 active topic
If 78% of US workers are at AI-using firms, why have so few jobs been officially lost?
Timeline for Survey of Business Uncertainty
Fed admits: no single AI adoption figure
AI: Jobs, Power & MoneyMentioned in: Tikhoretsk oil station struck twice
Russia-Ukraine War 2026- What does 78% AI adoption actually mean for US workers?
- The 78% SBU figure means 78% of employed Americans work at a firm that has adopted AI — not that 78% personally use AI tools. Daily individual use is 12%; weekly use is 35.2%.Source: Federal Reserve Board (FEDS Notes)
- Why does the government show both 18% and 78% AI adoption?
- The figures measure different things. BTOS counts firms (18%). SBU weights by employment (78%). The Federal Reserve published a reconciliation on 3 April 2026 showing all three federal instruments describe the same economy incompatibly.Source: Federal Reserve Board
Background
The Survey of Business Uncertainty (SBU) is a joint project of the Federal Reserve Bank of Atlanta, Stanford University, and the University of Chicago, measuring business expectations and economic uncertainty. Its late-2025 AI adoption finding — 78% of the US workforce is employed at firms that use AI — was the highest figure in the three-way comparison published by the Federal Reserve Board on 3 April 2026. The 78% figure is employment-weighted: it asks what share of the labour force works at a firm that has adopted AI, meaning a large AI-using employer counts for all of its workers.
The 78% figure does not mean 78% of workers personally use AI tools; it means 78% of employed Americans work inside an AI-using firm. The information sector leads at 37% firm adoption, with professional services and financial services each near 30%. Daily AI use across the full workforce sits at 12%; weekly use at 35.2%. The SBU employment-weighted methodology is the broadest measure because it captures workforce exposure to AI without requiring active individual use — a worker at a large bank using AI in its trading operations is counted even if that worker's own job has not changed.
For labour-market analysis the SBU's 78% is arguably the most important of the three federal figures, because AI's workforce impact runs through the firms workers are employed by — through changed hiring, changed job requirements, and changed wage premia — rather than through individual tool use alone. The Stanford Digital Economy Lab analysis, which found AI suppressing roughly 1 million annual hires, is implicitly consistent with a world where 78% of the workforce is inside AI-using firms but most have not yet seen their individual roles automated.