Summary
- The U.S. Census Bureau’s Business Trends and Outlook Survey (BTOS) replaced the term ‘Artificial Intelligence’ with ‘Super Intelligence’ (SI) from October 19, 2026, under Executive Order 14434.
- This is the second material question change in twelve months — the survey wording shifted in November 2025 from AI use ‘in producing goods or services’ to ‘in any business function’, which itself caused reported adoption rates to jump sharply.
- Large companies (250+ employees) went from roughly 12% AI adoption under the old ‘goods or services’ wording to around 40% by October 2026 under the broader ‘any business function’ framing — figures that are not directly comparable.
- The terminology change to ‘Super Intelligence’ introduces a fresh comparability problem; results under the new wording are due for publication imminently.
- Organisations benchmarking AI adoption using BTOS data should treat the pre-November 2025, post-November 2025, and post-October 2026 data series as three distinct datasets.
A survey in flux
Every two weeks, the U.S. Census Bureau surveys 200,000 businesses — drawn from a pool of 1.2 million — as part of its Business Trends and Outlook Survey (BTOS). Each business in the rotating panel completes the survey roughly every twelve weeks, and the latest results are published on alternating Thursdays. For many organisations tracking the pace of corporate AI uptake in the United States, this has been a primary data source.
Two question changes in twelve months
The BTOS has now changed its core AI question twice in under a year. Until November 17, 2025, businesses were asked whether they used AI ‘in producing goods or services’. On December 3, 2025, the Bureau announced a shift to the broader framing: ‘in any of its business functions’. That single wording change had a pronounced effect on reported adoption. Analysis of the BTOS data shows large companies — those with 250 or more employees — answered yes about 12% of the time under the old question. Under the new wording, the same cohort reported approximately 30% adoption by February 2026, rising to around 40% by October 2026. Those two figures are not measuring the same thing, and drawing a straight trend line between them is methodologically unsound.
Executive Order 14434 introduces ‘Super Intelligence’
Beginning with the October 19, 2026 collection cycle, the BTOS instrument changed again — this time driven by Executive Order 14434. The Bureau’s own about page confirms the shift: the term ‘Artificial Intelligence’ and the abbreviation ‘AI’ have been replaced throughout the survey with ‘Super Intelligence’ and ‘SI’. The revised question now reads: ‘In the last two weeks, did this business use Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI) in any of its business functions?’ The Bureau’s examples of SI use — including ‘generative SI for marketing, product design, or coding’ and ‘virtual SI agents for customer service’ — are identical in substance to the previous AI examples, with the acronym substituted throughout. The first published results under this new wording are expected shortly.
What this means for the data
The inclusion of the parenthetical ‘(formerly referred to as Artificial Intelligence or AI)’ is intended to help respondents make the connection, and may reduce confusion. However, businesses in the BTOS panel last answered the question up to twelve weeks ago, when the terminology was still ‘AI’. Whether the rebranding affects how respondents self-report — particularly smaller businesses less familiar with the policy nomenclature — remains to be seen. The Bureau has not yet published results under the new wording, so the impact on measured adoption rates is currently unknown.
Three series, not one
Security and strategy leaders who have used BTOS data to benchmark their organisation’s AI posture against industry peers now need to treat this as three distinct time series: adoption ‘in producing goods or services’ (pre-November 17, 2025); adoption ‘in any business function’ using AI terminology (November 2025 to October 2026); and adoption ‘in any business function’ using Super Intelligence terminology (October 2026 onwards). Conflating them will produce misleading trend lines. This matters practically when reporting to boards, justifying investment, or assessing competitive exposure.
Why it matters
CISOs and their peers frequently cite industry adoption benchmarks when making the case for AI governance frameworks, risk assessments, and technology investment. The BTOS has been one of the few publicly available, high-frequency datasets on corporate AI use in the United States. Two question changes in twelve months — the second driven by a federal executive order rather than methodological review — mean the headline figures circulating in board decks and analyst reports may be comparing incompatible data points. Any benchmarking exercise that treats the BTOS as a continuous series is now producing numbers that cannot be trusted at face value. Leaders should also note that regulatory terminology can shift in ways that affect compliance language, vendor contracts, and internal policy documents that reference ‘AI’ by name.
What to do now
- Audit any board or executive reports that reference U.S. BTOS AI adoption figures and flag the two question-change dates (November 17, 2025 and October 19, 2026) as series breaks before drawing trend conclusions.
- When the first BTOS results under the ‘Super Intelligence’ wording are published, compare them directly against the most recent ‘AI’ wording results rather than treating them as a continuation of the existing trend.
- Review internal AI policy documents, governance frameworks, and vendor contracts that use the term ‘Artificial Intelligence’ to assess whether a definitional shift in U.S. federal terminology — however it evolves — could affect compliance obligations or contractual scope.
- Do not use the jump from ~12% to ~40% large-company adoption as evidence of organic growth; document for stakeholders that the November 2025 question change is the primary driver of that increase.
