Business AI adoption accelerated before the survey changed
This chart tests whether business AI adoption accelerated before the Census survey question changed. It uses a data-selected piecewise trend on the consistently worded BTOS series, making the breakpoint exploratory rather than a confirmed structural change.
What does it show?
Under the original survey wording, the fitted adoption trend steepened from 1.2 to 4.4 percentage points per year around 2024-08-25.
Methodology
Uses only the consistently worded BTOS national series asking whether employer businesses used AI in producing goods or services. A continuous two-segment linear model searches eligible breakpoints with at least ten observations on either side and selects the lowest Bayesian Information Criterion. This is an exploratory descriptive breakpoint, not proof of a structural change; serial correlation and repeated breakpoint search make ordinary regression significance tests inappropriate.