{"id":660334,"date":"2026-10-14T10:53:44","date_gmt":"2026-10-14T08:53:44","guid":{"rendered":"https:\/\/preprod.aivancity.ai\/blog\/what-openais-study-reveals-about-chatgpt-broadening-jobs\/"},"modified":"2026-10-23T13:21:05","modified_gmt":"2026-10-23T11:21:05","slug":"what-openais-study-reveals-about-chatgpt-broadening-jobs","status":"publish","type":"post","link":"https:\/\/aivancity.ai\/blog\/en\/what-openais-study-reveals-about-chatgpt-broadening-jobs\/","title":{"rendered":"What OpenAI\u2019s Study Reveals About ChatGPT Broadening Jobs"},"content":{"rendered":"\n
\n

\n O<\/span>n September 16, 2026<\/strong>,\n the economic research<\/strong> team\n at OpenAI<\/strong>\n published a study covering\n more than 1.5 million work-related messages<\/strong>\n sent to ChatGPT<\/strong> by registered users\n in the United States<\/strong>.\n The findings suggest that some employees regularly\n return to tasks traditionally associated\n with other professions<\/strong>.\n This shift may\n broaden the content of jobs<\/strong>\n even before their titles change,\n but the study measures neither\n the productivity achieved, nor the quality of the results,\n nor the effects on employment and wages.<\/strong>\n <\/p>\n<\/div>\n\n

\n\n
\n 01\n <\/div>\n\n

\nA study of how ChatGPT is actually used at work\n <\/h2>\n\n<\/div>\n\n

The Work at the Frontier report analyzes more than 1.5 million messages classified as work-related, sent between April and July 2026 from individual ChatGPT accounts. The participants also had a ChatGPT Business account, which allowed OpenAI to link their messages to an occupation or department declared at sign-up.[1]<\/a><\/p>\n\n

This methodological detail matters. The messages do not come from companies\u2019 ChatGPT Business workspaces, and the study does not observe all of the work performed. OpenAI says it excluded users who had opted out of having their data used for training, automatically removed personal information before classification, and prevented researchers from reading individual conversations.[1]<\/a><\/p>\n\n

The lab then compares the requests with the activities listed in O*NET, the US database of occupations and tasks. A request is considered \u201ccross-functional\u201d when it corresponds more closely to another occupational family than to the occupation declared by the user. The study thus seeks to determine whether AI merely enables one-off experimentation or whether certain new activities become part of everyday routines.<\/p>\n\n

\n\n
\n 02\n <\/div>\n\n

\nTasks outside one\u2019s occupation that become recurring\n <\/h2>\n\n<\/div>\n\n

Among roughly 6,200 people tracked continuously over the four months, the share of messages corresponding to cross-functional tasks they had already used rose from 13.1% in April to 25.9% in July. The indicator nearly doubled, but the way it is built mechanically favors an increase: each new month expands the number of tasks that can qualify as recurring. OpenAI points this out in the report.[1]<\/a><\/p>\n\n

A second comparison nonetheless strengthens the hypothesis that these practices are taking hold. When a user had turned to ChatGPT for a cross-functional task in a given month, they returned to that same category the following month in 23.6% of cases, compared with 8.4% among comparable users who had not used it before. That is a gap of 15.2 percentage points.[1]<\/a><\/p>\n\n

\n \n \n \n \n \n \n \n \n \n
\n Observed indicator\n <\/th>\n\n \n Published result\n <\/th>\n\n \n What it tells us\n <\/th>\n <\/tr>\n <\/thead>\n\n
\n Messages analyzed\n <\/td>\n\n \n More than 1.5 million\n <\/td>\n\n \n A large sample of work-related requests, but limited to a selected population of US users.\n <\/td>\n <\/tr>\n\n
\n Cohort tracked for four months\n <\/td>\n\n \n About 6,200 users\n <\/td>\n\n \n Analysis of recurrence among people with at least ten sampled messages each month.\n <\/td>\n <\/tr>\n\n
\n Share of cross-functional tasks already used\n <\/td>\n\n \n 13.1% in April; 25.9% in July\n <\/td>\n\n \n An increase consistent with the integration of new tasks, without demonstrating a contractual change in the job.\n <\/td>\n <\/tr>\n\n
\n Return the following month\n <\/td>\n\n \n 23.6% vs. 8.4%\n <\/td>\n\n \n A task already performed with AI reappears more often, but the relationship alone does not establish causality.\n <\/td>\n <\/tr>\n\n
\n Average return after first use\n <\/td>\n\n \n 18.5%\n <\/td>\n\n \n About one in five user-task combinations is reused the following month.\n <\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n<\/div>\n\n
\n\n
\n 03\n <\/div>\n\n

\nNot all task expansions are equal\n <\/h2>\n\n<\/div>\n\n

Return rates vary widely depending on the nature of the work. Customer interactions reach 54%, advertising or promotional copywriting 44%, and the creation of marketing materials 37%. By contrast, legal research is repeated in only 10% of cases, while explaining financial information and presenting business information reach 15%.[1]<\/a><\/p>\n\n

This difference may reflect how often tasks naturally occur, but also the level of perceived risk. Producing a first draft of a sales message is relatively easy to check. Interpreting a legal rule or explaining financial information requires expertise, validation and more clearly defined accountability. A low repetition rate therefore does not mean that AI is incapable of helping, but that its integration depends on the context and on the consequences of an error.<\/p>\n\n

\n \n\n \n \n \n\n \n \n \n \n \n
\n Cross-functional task\n <\/th>\n\n \n Return rate the following month\n <\/th>\n\n \n Cautious interpretation\n <\/th>\n <\/tr>\n <\/thead>\n\n
\n Discussing goods or services with customers\n <\/td>\n\n \n 54%\n <\/td>\n\n \n Frequent, easily integrated use, provided the information shared is checked.\n <\/td>\n <\/tr>\n\n
\n Writing advertising or promotional content\n <\/td>\n\n \n 44%\n <\/td>\n\n \n Generation speeds up first drafts but guarantees neither compliance nor brand consistency.\n <\/td>\n <\/tr>\n\n
\n Creating marketing materials\n <\/td>\n\n \n 37%\n <\/td>\n\n \n AI makes production easier without replacing market knowledge and human validation.\n <\/td>\n <\/tr>\n\n
\n Explaining financial information\n <\/td>\n\n \n 15%\n <\/td>\n\n \n The risk of error and accuracy requirements may limit delegation.\n <\/td>\n <\/tr>\n\n
\n Conducting legal research\n <\/td>\n\n \n 10%\n <\/td>\n\n \n Low recurrence may reflect the need for up-to-date sources and specialized skills.\n <\/td>\n <\/tr>\n\n <\/tbody>\n <\/table>\n<\/div>\n\n
\n\n
\n 04\n <\/div>\n\n

\nDoes AI really let people borrow expertise?\n <\/h2>\n\n<\/div>\n\n

OpenAI observes that cross-functional requests are slightly shorter, by about twenty characters on average. They less often ask for explanations, advice or a detailed format, but they provide more examples and context and more frequently request verification. The authors put forward the idea of \u201cborrowed expertise\u201d: the user brings a problem and contextual elements, then relies on the model to apply knowledge from another field.[1]<\/a><\/p>\n\n

This interpretation calls for caution. A well-worded output does not prove that a skill has been acquired. Users can obtain a deliverable without understanding the method, the assumptions or the limitations. In the short term, this reduces friction between departments. In the longer term, organizations must distinguish assisted autonomy, which relies on the ability to check the work, from uncontrolled delegation that could weaken learning.<\/p>\n\n

The International Labour Organization estimates that one in four workers worldwide holds a job with some degree of exposure to generative AI. It nevertheless considers the transformation of tasks more likely than the complete disappearance of most jobs, notably because human intervention remains necessary.[2]<\/a> OpenAI\u2019s study offers a behavioral perspective consistent with this view, without directly measuring employment.<\/p>\n\n