Skip to content
Briefings are running a touch slower this week while we rebuild the foundations.See roadmap
AI: Jobs, Power & Money
1JUN

Fed admits: no single AI adoption figure

4 min read
09:18UTC

The Federal Reserve Board published a reconciliation paper on 3 April comparing three federal instruments that describe the same late-2025 economy as 18%, 41% or 78% AI-adopted. The gap is structural.

← Back to Markets now reward the cut, punish the freezeJump to analysis ↓
EconomicDeveloping
Key takeaway

Federal agencies disagree by 4.3x on AI adoption, leaving displacement policy without a statutory baseline.

The Federal Reserve Board published a survey reconciliation paper on 3 April 2026, authored by staff economist Jeffrey S. Allen for the FEDS Notes series, comparing three separate federal instruments that should describe the same US economy and do not. The Business Trends and Outlook Survey (BTOS), run by the Census Bureau on a firm-weighted basis, put AI adoption at 18% for late 2025. The Research on Practices Survey (RPS), which asks individuals whether they personally use AI at work, returned 41%. The Survey of Business Uncertainty (SBU), which weights by how many workers are employed at AI-using firms, came in at 78%. Daily AI use across the US workforce sits at 12%; weekly use at 35.2%.

The three figures measure different things: the share of firms using AI, the share of workers personally using AI, and the share of the workforce employed at firms that have adopted it. The Fed paper's point is not that the surveys are faulty; it is that no federal agency has formally chosen between firm-weighted, individual-weighted and employment-weighted units, so the same quarter can be described as roughly one-in-five, two-in-five or four-in-five AI-adopted depending on which federal instrument is cited.

Forty-six days earlier, the bipartisan nine-senator coalition led by Josh Hawley and Mark Warner had written to the Department of Labor and the Bureau of Labor Statistics urging expanded AI workforce data collection . The Fed Board's reconciliation is effectively the federal answer: there is no single figure, and none of the three existing instruments will produce one until an agency chooses a canonical unit. The BLS itself has so far chosen none, skipping a separately scheduled GenAI workplace publication eleven days later.

In practice, private datasets set the headline number by default. Challenger, Gray & Christmas counts announced AI layoffs; Stanford's JOLTS-based analysis puts the real labour impact at a far higher multiple; Goldman Sachs's earlier monthly substitution model sat between the two. None of those are federal. None can be legislated against without a statutory benchmark that the reconciliation paper has just confirmed does not exist.

The 4.3x divergence is also the quantitative foundation for the Hawley-Warner demand: a letter asking for one number has been met by a paper documenting that the government currently produces three, none of them designated as authoritative. Whether the BLS is resourced to build a fourth, harmonised instrument, or whether the NY Fed's Survey of Consumer Expectations becomes the de facto federal measure by attrition, is likely the first concrete thing the next US Congress will have to decide on AI workforce policy.

Deep Analysis

In plain English

The US government tried to measure how many workers are using AI at work and got three completely different answers: 18%, 41%, and 78%, all for the same period. Each answer came from a different official survey asking a slightly different question. This matters because the same data is used to set policy on jobs, training budgets, and unemployment support. Right now, no one in government can agree on the most basic fact: how widely AI has spread through the American workplace.

Deep Analysis
Root Causes

The 4.3x gap traces to a structural decision each agency made independently: BTOS asks whether a firm has 'adopted' AI (binary, firm-level), RPS asks whether an individual 'uses' AI tools (self-report, individual-level), and SBU weights by employment rather than firm count. No interagency protocol required these to be compatible before deployment.

The deeper structural condition is that the US federal statistical system was designed in an era of physical goods and discrete employment events. AI is neither; it is a probabilistic workflow layer whose adoption is continuous, partial, and contested even within firms. The BLS's Occupational Employment Statistics programme, which underpins most workforce policy, still classifies jobs by task bundles last revised in the O*NET 2010 review cycle.

Sources:Federal Reserve Board (FEDS Notes, Jeffrey S. Allen)·Federal Reserve Board (FEDS Notes, Jeffrey S. Allen)
Mentions:Federal Reserve →Business Trends and Outlook Survey →Research on Practices Survey →Survey of Business Uncertainty →Bureau of Labor Statistics →US Congress →Census Bureau →Goldman Sachs →Josh Hawley →Hamilton Project →Survey of Consumer Expectations →Challenger, Gray & Christmas →Senator Mark Warner →Job Openings and Labor Turnover Survey →
First Reported In

Update #6 · Three federal surveys, one 34-to-1 gap

Federal Reserve Board (FEDS Notes, Jeffrey S. Allen)· 16 Apr 2026
Read original →
Causes and effects
Caused by
Nine senators demand AI workforce data
The nine-senator Hawley-Warner letter demanding AI workforce data from federal agencies directly prompted the Fed Board to publish its 3 April reconciliation paper documenting the 4.3x measurement gap.
Occurred 1 Mar 2026
Read story →
This Event
Fed admits: no single AI adoption figure
This is the first federal acknowledgement that no canonical AI adoption measure exists. Without an agreed baseline, Congress cannot legislate against a figure, and the Hawley-Warner measurement coalition retains political legitimacy as the only realistic path to better data.
Led to
UKG cuts 6%; Blackstone's AI playbook
NY Fed SCE finding that 62% of employed Americans expect AI to increase unemployment within twelve months is the sentiment backdrop against which UKG's cuts were made.
Occurred 15 Apr 2026
Read story →
Two federal studies split on AI jobs
The Fed Board's three-survey reconciliation showing 18%, 41% and 78% adoption for the same period compounds the NY Fed's methodological split into a federal-measurement crisis.
Occurred 14 May 2026
Read story →
Different Perspectives
TSMC and Taiwan chip supply chain
TSMC and Taiwan chip supply chain
Nvidia's 17% headcount growth to 42,000 on $81.6 billion in quarterly revenue depends on TSMC's CoWoS advanced packaging capacity constraining H100 and B200 supply, sustaining margins above 70%. The AI build-out's sole headcount-growth story runs through a Taiwan supply chain that has no parallel in downstream software.
Displaced tech workers globally
Displaced tech workers globally
CrowdStrike's SEC disclosure puts AI attribution on a material regulatory record for the first time, but Oracle's Massachusetts WARN clock expired unfiled after up to 14 workers were logged as remote despite office proximity. The legal apparatus cannot enforce what it cannot see: hybrid reclassification, GCC transfers, and hires never made.
UK workforce and policymakers
UK workforce and policymakers
ONS recorded UK vacancies at 705,000, below the pre-pandemic baseline for the first time, as payrolled employment fell 210,000 year on year with real wage growth at 0.1%. The Bank of England's AI worst case assumed 500,000 additional unemployed from a baseline above 730,000; the UK is already below that floor, and ONS still publishes no AI-exposure breakdown.
India IT workforce and graduates
India IT workforce and graduates
NASSCOM's FY2026 data shows net sector growth of 140,000, but entry-level hiring fell 20-25% as the growth concentrated in in-house GCC offices requiring mid-career specialists. Indian graduates who previously entered through TCS, Infosys and Wipro fresher programmes find that channel closing at both ends: outsourcers cutting and GCCs not hiring at the junior level.
IG Metall and European trade unions
IG Metall and European trade unions
European labour bodies see the market reward pattern, cuts on record revenue, as investor preference for short-term margin extraction over validated AI productivity. They note the EU Digital Omnibus provisional deal has dropped binding employer AI-literacy obligations at the precise moment the ILO-NASK index has quantified that 3.3% of global workers are in the highest AI exposure category.
Federal Reserve Board
Federal Reserve Board
Governor Cook told Stanford's SIEPR on 27 May that speculative-grade software bond spreads have widened on AI-disruption concern, moving AI displacement from a labour observation into the Fed's financial-stability mandate. The Fed cannot resolve structural labour transformation through rate policy, so Cook routed the concern through the one channel the Fed does control.