According to a November 2, 2026 post by Simon Willison, the U.S. Census Bureau's Business Trends and Outlook Survey (BTOS) has changed the wording of its AI-related question, most recently to comply with Executive Order 14434. The BTOS surveys 200,000 of 1.2 million businesses every two weeks, with each business cycling through roughly every twelve weeks, and publishes data on alternating Thursdays. Willison reports two material changes. First, on a date he identifies around November 17, 2025 (announced December 3, 2025), the question shifted from asking whether a business used AI 'in producing goods or services' to 'in any of its business functions.' Second, beginning with the October 19, 2026 collection, the instrument replaced 'Artificial Intelligence (AI)' with 'Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI).' Willison notes these wording changes mean responses before and after the changes are not directly comparable. Using the October 22, 2026 data, he reports that large companies (250+ employees) answered 'yes' around 12% in September 2025 under the old wording, about 30% by February 2026 under the broader wording, and roughly 40% by October 2026. The first results reflecting the 'Super Intelligence' wording were due the following Thursday.
- Survey wording changed from AI 'in producing goods or services' to 'in any of its business functions' around November 17, 2025
- Beginning October 19, 2026, the term was changed to 'Super Intelligence or SI' due to Executive Order 14434
- Reported 'yes' responses from large firms (250+ employees): ~12% Sept 2025, ~30% Feb 2026, ~40% Oct 2026
- Wording changes make before-and-after responses not directly comparable, per Willison
What it means for you
If you rely on national statistics to judge how fast your peers are adopting AI, be careful: the Census Bureau changed what its headline survey question actually asks twice in one year. That means the jump from ~12% to ~40% adoption among large firms is partly a measurement artifact, not purely real-world growth — a broader question catches more 'yes' answers. Treat any 'AI adoption is surging' or 'slowing' chart built on this survey with caution.
Try this
Next time someone shows you an adoption statistic to justify a decision, ask one question: what exactly was being measured, and did the definition change? That habit catches a lot of misleading numbers before they shape a budget.
Who should care
Owners and operations leads who use industry benchmarks or adoption stats to decide whether they're behind on AI, and anyone building a case for or against new tools on 'everyone else is doing it' data.
Skip this if
You don't make decisions based on national adoption surveys and don't cite them — then this is just an amusing footnote about a renamed government question.
Sources: Simon Willison — read the original