Skip to content
Future Frontpage FF markFFFUTUREFRONTPAGE

The AI developments that belong on page one.

Foundation models

Census AI adoption data spikes after survey wording shift

A change in the U.S. Census Bureau’s survey question caused a jump in reported AI usage among large firms, according to analysis by Simon Willison.

Key points

  • Large firms reported 12% AI usage in Sept 2025 under the old "producing goods or services" question.
  • Usage rose to 30% by Feb 2026 and 40% by Oct 2026 under new "any business function" wording.
  • The Bureau announced the question change on Dec 3, 2025, following the Nov 17 shift.
Census AI adoption data spikes after survey wording shift
Depth
12%Large firms reported 12% AI usage in Sept 2025 under the old "producing goods or services" question · reported
30%Usage rose to 30% by Feb 2026 and 40% by Oct 2026 under new "any business function" wording · reported
40%Usage rose to 30% by Feb 2026 and 40% by Oct 2026 under new "any business function" wording · reported
1 outlet · newest source 49h ago

The U.S. Census Bureau’s data shows a sharp rise in AI adoption among large companies after a survey question change, according to analysis by Simon Willison.A change in how the Census Bureau asks about AI use caused reported adoption numbers to jump significantly.The semantic shift in BTOS questioning from production-specific to general business functions inflated reported AI adoption metrics.

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

The figures

Apollo Global Management published a chart in September 2025 showing U.S. corporate AI adoption trends based on Census Bureau data. Willison’s updated chart uses the most recent numbers published on October 22, 2026. In September 2025, large companies with 250 or more employees answered "yes" about 12% of the time to the question regarding AI in producing goods or services. By February 2026, under new wording, those same companies answered "yes" around 30% of the time. By October 2026, that figure increased to around 40%.

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Behind the figures

The U.S. Census Bureau runs the Business Trends and Outlook Survey (BTOS) every two weeks. Each survey reaches 200,000 businesses out of a total of 1.2 million. The panel cycles so an individual business completes the survey every twelve weeks. Data is published on alternating Thursdays at 10am. The Bureau announced a significant change to the survey question on December 3, 2025. The question had changed materially on November 17, 2025.

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

What it means for your stack

The previous question asked if businesses used AI in producing goods or services. Examples included machine learning, natural language processing, virtual agents, and voice recognition. The new question uses "any business function" wording, according to the published report description of the chart update. This broader scope captures AI uses beyond direct production. Willison noted the Bureau has fallen foul of Executive Order 14434, also known as the Super Intelligence order.

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Evidence: reportedThe U.S. Census Bureau and “Super Intelligence”

aifleet-willison

Building on this

Developers can build dashboards that track these shifting metrics by parsing the Bureau’s Employment size class XLSX file. The data provides a weekly pulse on corporate AI integration. Engineers can create visualizations that compare the old production-focused metrics with the new business-function metrics. This allows for analysis of how broader definitions change perceived adoption rates.

Why it matters

  • The definition of AI usage in official statistics has broadened significantly, affecting benchmark comparisons.
  • Large firms report higher AI usage when the metric includes any business function, not just production.
  • The survey panel size of 200,000 businesses provides a substantial dataset for trend analysis.
  • Weekly publication cycles allow for near-real-time monitoring of corporate technology shifts.

Topics: census data, AI adoption, survey methodology, corporate trends

Updated

For readers

Developers

  • Parse the Employment size class XLSX file to build adoption dashboards.
  • Track weekly BTOS releases for real-time corporate AI trend data.

Policy

  • Note the impact of survey wording changes on reported adoption metrics.
  • Consider how broader definitions affect regulatory and funding decisions.

Researchers

  • Compare pre- and post-November 2025 data for methodological shifts.
  • Analyze the correlation between Executive Order 14434 and survey changes.

Numbers in context

12%reported30%reported40%reportedCircle area is proportional to the magnitude; labels retain signs. Figures may describe different measures. Read their contexts above.

Ask the article

Local keyword search of this article; no generated answers.

Did this article give you the evidence you needed?

No vote saved on this device.
Your choice stays on this device. No public tally.

Questions readers ask

How did the Census Bureau survey question change?

The question shifted from asking about AI in producing goods or services to any business function.

What was the AI adoption rate for large firms in September 2025?

Large firms with 250+ employees reported AI usage in producing goods or services around 12% of the time.

When did the Bureau announce the survey change?

The Bureau announced the change on December 3, 2025, after the question shifted on November 17.

Connected stories

Reading receipt

  • Large firms reported 12% AI usage in Sept 2025 under the old "producing goods or services" question.
  • Usage rose to 30% by Feb 2026 and 40% by Oct 2026 under new "any business function" wording.
  • The Bureau announced the question change on Dec 3, 2025, following the Nov 17 shift.

Still open: What is the exact full text of the new survey question?

Still open: How does Executive Order 14434 specifically affect Census Bureau operations?

Time spent: 0:00 · Depth: Full

Sources

  1. Simon Willison AI

Related stories