- thekb.eu
- Knowledge graph
- Document
- Work at the Frontier
Work at the Frontier
Work at the Frontier — Document. reference: OpenAI Economic Research, "How AI is Expanding What People Do at Work," 1st report in the series, July 27, 2026 — over 800,000 messages from US ChatGPT users
Work at the Frontier is a report series from OpenAI Economic Research whose first instalment, «How AI is Expanding What People Do at Work», appeared on 27 July 2026 and draws on more than 800,000 messages from American ChatGPT users. It is built on the «AI Jobs Transition Framework», and OpenAI Economic Research puts a single measurement at its centre: task crossover, work historically associated with one occupation showing up in the AI use of people in another.
Two numbers describe the same phenomenon. 16.8% of work-related messages concern a task belonging to another occupation; among occupation-specific messages the share is 43.5%. The gap is a funnel: 61.5% of professional use is generic (writing, summarising, planning) and set aside before the ratio is computed.
The per-function split is the sharpest part. Customer experience 77%, design 75%, HR 69%, legal 56%, marketing 53%, sales and finance 40%, engineering 28%. Flows run in opposite directions: design imports 35.2% of outside tasks and exports 1.7%, engineering does the reverse at 18.5% and 7.4%, marketing does both (24.3% in, 8.9% out, the largest outgoing share in the sample). On the heatmap diagonal, engineering retains 53% of its own work while customer experience retains 11% and HR 10%.
The size effect is thinner than it looks: the out-of-occupation share falls from 18.9% to 16.3% with workspace size among average users only. OpenAI Economic Research claims the status of a leading indicator, readable before firms rewrite job descriptions. The data measures OpenAI's own product, on American users alone.
- Type
- Document
- reference
- OpenAI Economic Research, "How AI is Expanding What People Do at Work," 1st report in the series, July 27, 2026 — over 800,000 messages from US ChatGPT users
- relations
- 15
- Cited in
- 2 fiches
Adoption measures
-
16,8% of work-related messages and 43,5% of occupation-specific messages concern a task from another occupation
-
out-of-occupation share by function: customer experience 77%, design 75%, HR 69%, legal 56%, marketing 53%, sales 40%, finance 40%, engineering 28%
-
43,5% of occupation-specific messages fall outside the user's occupation
-
61,5% of professional usage is generic and excluded from the crossover calculation
-
engineering imports 18,5% of external tasks and exports 7,4%
-
the out-of-occupation share goes from 18,9% to 16,3% depending on workspace size, among average users only
-
design imports 35,2% of external tasks and exports only 1,7%
-
marketing imports 24,3% and exports 8,9%, the highest outbound share in the sample
Neighborhood
← references
← publishes
→ is based on