AI use is broadening; the employment evidence needs a separate reading

Among 74 countries observed throughout Q1 2025–Q3 2026, the average share concentrated in the ten most common activities fell from 56.0% to 51.1%. The mix is broadening, but these are classified messages on individual plans; enterprise use and Codex are excluded. The release measures neither productivity nor employment effects. [1][2]

Enterprise access is expanding separately. Uber’s October 22 announcement begins a ChatGPT Work rollout to more than 30,000 employees, following a 1,000-person pilot. Training, approved integrations and an internal skills library accompany access. The announcement does not report sustained active use, time savings or a controlled quality comparison; availability is the measured rollout milestone. [3]

Pew’s October 22 analysis finds a larger male–female employment-growth gap in occupations with greater AI exposure. It uses about four million Current Population Survey records across 480 occupation groups, comparing 2017–19 with January 2023–June 2026 and excluding the pandemic years 2020–22. Its task-based exposure score describes potential AI assistance, not each employer’s actual adoption. The association cannot on its own identify AI as the cause. [4][5]