Measured adoption of AI by firms

Assembled from 26 blocks · 1 evidence · 24 related

1. Story
2. What organisations actually do, as opposed to what systems can do — and a case study in how much a measurement depends on the question asked. The United States Census Bureau's Business Trends and Outlook Survey has asked firms about AI use fortnightly since 2023, and under its original wording — use "in producing goods or services" — the rate rose from 3.7% in September 2023 to 5.4% by February 2024 and reached roughly 10% by late 2025. On 17 November 2025 the question was rewritten to ask about use "in any of its business functions". The reported national rate immediately stood at 17.3%, and through the first half of 2026 ran between about 17% and 20%, with adoption concentrated in larger firms — around 37% among businesses with at least 250 employees in the period ending 3 May 2026. Nothing about firm behaviour changed on 17 November 2025. The instrument changed, and a reader who compares a 2026 figure with a 2024 figure across that break is measuring the questionnaire.
3. Knowledge
4. Benchmark saturation
5. Measured adoption of AI by firms
6. Labour-market effects of machine learning
7. Connections
8. Labour-market effects of machine learning
9. Benchmark saturation
10. A whole-labour-market analysis finds no economy-wide disruption yet
11. The United States firm-adoption survey changes its question, and the number jumps
12. Machine translation
13. Machine-generated code
14. Measured adoption of AI by firms
15. Payroll data shows a relative decline in entry-level employment in AI-exposed occupations
16. A whole-labour-market analysis finds no economy-wide disruption yet
17. The second International AI Safety Report reports mixed labour findings
18. Radiologist workforce projections and employment outlook, against the 2016 prediction
19. The difficulty of evaluating machine learning systems
20. Measured adoption of AI by firms
21. Fei-Fei Li
22. GLUE is saturated and SuperGLUE is built to replace it
23. An audit finds the most-cited preference leaderboard is optimised rather than merely measured
24. The AI Index records both fast benchmark movement and doubts about benchmarks
25. Evidence
26. Under the original wording — AI use "in producing goods or services" — the fortnightly BTOS AI use rate rose from 3.7% in September 2023 to 5.4% in February 2024 and reached roughly 10% by late 2025. On 17 November 2025 the question was rewritten to ask about use "in any of its business functions", with the supplement expanded to fifteen functions, and the reported national rate stood at 17.3%; through December 2025 to May 2026 the rate ran between about 17% and 20%, with about 37% among firms of at least 250 employees and 32% among firms of 100–249 employees in the period ending 3 May 2026. V55 verification basis: no Census page was fetched. The early series, the 17 November 2025 rewording, the 17.3% figure, the 17–20% band and the firm-size breakdown were carried in search results attributing them to Census BTOS releases and documentation and to the St. Louis Fed commentary on the wording change.