Task Crossover: OpenAI's Work at the Frontier Data

Key takeaways

1. Date / source. July 27, 2026, OpenAI Economic Research, first installment in the Work at the Frontier series, based on more than 800,000 messages from US ChatGPT users.
2. Key framing. the paper doesn't ask whether the model can perform a fixed list of tasks, but who does what — « Our evidence suggests that AI is also changing who takes on which tasks. » It measures an observed redistribution, not a theoretical substitutability. ### The two headline figures, always to be given together 16.8% of work-related messages and 43.5% of occupation-specific messages. The funnel: 61.5% of professional usage is generic (writing, summarizing, planning — « shared too broadly across occupations to be evidence of crossover ») and removed from the calculation; of the remaining 38.5%, 43.5% fall outside the occupation. The 43.5% is thus calculated on 38.5% of usage, and citing it alone nearly doubles the impression of scale. Additional nuance: the complement is not "within the occupation" but « inside or near a user's occupation » — the measured boundary is fuzzy by construction. ### By occupation, outside-occupation share Customer experience 77% · design 75% · HR 69% · legal 56% · marketing 53% · sales 40% · finance 40% · engineering 28%. « Outside occupation work is a majority in five of eight groups. » ### The heatmap, the richest object in the report Rows = user's occupation, columns = task's occupation of origin; diagonal in bold. | User ↓ / Task → | CX | Design | Eng. | Fin. | HR | Legal | Mktg | Sales | |---|---|---|---|---|---|---|---|---| | Customer experience | 11 | 5 | 20 | 14 | 6 | 6 | 26 | 11 | | Design | 6 | 12 | 28 | 10 | 4 | 4 | 28 | 8 | | Engineering | 4 | 4 | 53 | 9 | 4 | 4 | 20 | 2 | | Finance | 6 | 4 | 22 | 23 | 4 | 8 | 25 | 7 | | HR | 9 | 5 | 18 | 16 | 10 | 10 | 23 | 9 | | Legal | 5 | 4 | 17 | 12 | 7 | 31 | 18 | 6 | | Marketing | 10 | 6 | 17 | 11 | 4 | 4 | 36 | 12 | | Sales | 11 | 5 | 18 | 15 | 4 | 6 | 29 | 12 | Three readings the text doesn't offer. The diagonal is the real result: engineering retains 53% of its own work, customer experience 11%, HR 10%, design 12%, sales 12%. For four occupations, the marketing column exceeds the diagonal — a designer does more marketing tasks (28%) and engineering tasks (28%) than design tasks (12%). Two columns absorb everything: marketing (18 to 36% everywhere) and engineering (17 to 53%). Caveat: a low diagonal doesn't mean these employees no longer do their own job, but that their non-generic messages are classified elsewhere, on a base that already excludes 61.5% of usage. The tasks specific to these occupations may be precisely the ones classified as "generic." ### Directions of circulation | Occupation | Imports | Exports | |---|---|---| | Design | 35.2% | 1.7% | | Engineering | 18.5% | 7.4% | | Marketing | 24.3% | 8.9% — « the highest outward share in the sample » | Marketing is the only one that does both. Two tasks are universally borrowed — financial calculation and technology troubleshooting appear in the top 3 of borrowings for the other seven groups; marketing-material creation appears among five other groups. ### The asymmetry of engineering Engineering is both the least porous occupation (28%, imports 18.5%) and a major exporter (7.4%), with the highest diagonal (53%). The celebrated porosity is thus largely one-way toward the technical side: other occupations do troubleshooting and technical-systems work, the reverse is rare. AI doesn't dilute engineering into other occupations, it spreads engineering tasks everywhere else while leaving engineers on their core work. ### The size effect is weaker than reported elsewhere The shift from 18.9% to 16.3% holds « among average users » only, and « among the heaviest users, we do not see the same monotonic pattern ». OpenAI offers two possible explanations — moderate users at small organizations turn to AI when they encounter work belonging to another function; heavy users have stabilized workflows or use AI more intensively within their core occupation. The conclusion is conditional: « AI may be especially useful as a generalist tool where specialist resources are scarce. » Correction to [[sfeir-ia-frontieres-metiers-skill-based-organisation-2026-08-01]], which treats these 2.6 points as a clean result and the generalist conclusion as an assertion: the source is more cautious than its commentary. ### Claimed epistemic status An early signal: this data makes it possible to see recombinations « before firms rewrite job descriptions or create new job titles » and « may provide an early signal of occupational change that conventional labor-market statistics will capture only later ». A leading indicator, not an employment measure — it says what people are attempting, not what organizations have ratified. The paper draws on the same team's AI Jobs Transition Framework, whose thesis is that many occupations will be reorganized rather than disappear. ### Underlying caveats
3. OpenAI measures OpenAI. — a single assistant, its own users only, claimed as an advantage (« our unique window »). What is measured is ChatGPT usage, not AI usage.
4. US users only. , a self-selected population not representative of the workforce.
5. The occupation classification is done by a model. , on messages, with no external validation described. The eight categories and the generic/specific boundary are design choices that mechanically determine the results — this is where the robustness of the 43.5% is actually decided.
6. Stated aim. « to guide policy and practice ». An actor that produces the data on its own impact and aims to shape public policy.