Gallup tracks what workers say about their own AI use

The setup

Gallup runs a recurring workplace tracking poll through its Gallup Panel, a self-administered web survey of a random sample of U.S. adults working full or part time, fielded quarterly. The second-quarter-2026 wave, reported in a Gallup Workplace article, ran from 6 to 20 May 2026 among 22,573 employed U.S. adults, with a stated margin of error of plus or minus 0.9 percentage points at the 95% confidence level.

What the documents show

That wave found 47% of employees reporting some organizational AI integration and 52% saying they personally use AI in their role, with 30% using it a few times a week or more and 15% daily. Among users, the most common reported uses were writing or editing (51%), search or research (49%) and general problem-solving (39%), with coding assistance and workflow automation each reported by 16%. Every one of these is what a respondent told Gallup about their own behavior, not a count drawn from any product's server logs.

The friction

That self-reported nature is the poll's central limit: a worker may over- or under-estimate how often they use AI, may define using AI differently than a colleague using the same tool inside a different app, and the poll cannot see whether reported use was for a genuine work task or a passing experiment. Gallup's own design-effect adjustment, stated alongside the release, signals the panel is weighted rather than a raw simple random sample, a technical detail easy to drop when the topline number gets quoted elsewhere.

What changed in the work

What a large, quarterly worker-side panel adds, compared with an employer survey like McKinsey's or a vendor's login count, is the employee's own perspective on frequency and specific use cases, tracked consistently across quarters with a published margin of error. It is a different vantage point on the same underlying question, not a more or less authoritative one, and conflating a self-reported usage rate with a logged-usage figure from a vendor's own product overstates what either source alone can show.

• Is a workplace-AI figure measuring what employees say they do, or what a system recorded them doing?
• Does the cited wave give a margin of error and sample size, or just a topline percentage?
• Would 'using AI' in this survey's wording match how the reader's own workplace defines the term?

Because this is a tracking poll rather than a one-time survey, its real value sits in the wave-over-wave change Gallup itself reports, such as the six-point rise in organizational integration cited between quarters, rather than in any single quarter's level read in isolation.