Task crossover: What the OpenAI report means for SMEs
Norwegian SME leaders already handle generalist work themselves. A new report from OpenAI now confirms that this is common in much larger organizations as well.
In the Work at the Frontier report, OpenAI Economic Research analyzed more than 800,000 work-related ChatGPT messages from business users. The main finding is that 16.8 percent of all tasks cross professional boundaries. Looking only at job-specific tasks, the share is 43.5 percent. In other words, almost half of the specialized tasks people perform in ChatGPT actually belong to a different occupation.
For Norwegian SMEs with 7-100 employees, this is hardly news. It is everyday reality.
What did OpenAI measure?
OpenAI first removed the generic tasks, such as writing, summarizing, and planning, which account for 61.5 percent. The remaining 38.5 percent were classified according to O*NET, the US Department of Labor's task database, according to Tech Times' review of the method. The researchers then checked whether the task code belonged to the user's own occupation or another one.
The result varies widely between professional fields:
- Customer experience: 77 percent crossover
- Design: 75 percent
- HR: 69 percent
- Legal: 56 percent
- Marketing: 53 percent
According to Axios' coverage of the report, OpenAI Chief Economist Ronnie Chatterji says that the boundaries between occupations are probably already becoming more fluid because of AI. The report puts it this way: "AI may be especially useful as a generalist tool where specialist resources are scarce."
What does this mean for Norwegian SMEs?
Norwegian SMEs are often run by generalists. An accountant also writes customer letters and publishes on LinkedIn. The managing director of an agency with 12 employees reviews contracts, troubleshoots websites, and analyzes suppliers without bringing in a lawyer, chief financial officer, or developer.
The OpenAI figures show that this pattern also exists in larger companies in the US sample. Small units with 2-5 employees have an 18.9 percent out-of-occupation share, while large ones with 100 or more employees are at 16.3 percent. The difference is smaller than one might have expected.
For a leader, this means three things:
- You no longer compete solely with other SMEs. You compete with larger organizations that can now perform more tasks internally with the help of AI.
- Generalist skills are becoming more important. Someone who can formulate a legal question, a design brief, or an HR policy with AI as a sparring partner can handle more types of tasks. The result still needs quality assurance from someone with the right professional expertise.
- Tasks that cross professional boundaries are worth mapping before an automation project. Start with repetitive, low-risk tasks.
What does the report not measure?
It is easy to get carried away. The OpenAI report measures behavior, not employment. It does not say whether AI creates new cross-functional work or simply helps with existing tasks. It says nothing about productivity, quality compared with a specialist, hiring, or wages.
Digital Applied highlights two weaknesses: the report was written by the vendor itself, and the classifier that placed the tasks into O*NET categories is OpenAI's own. Two different denominators are used for the same phenomenon: 16.8 percent of all tasks and 43.5 percent of job-specific tasks. Without a common denominator, that is only half the picture.
Jed Kolko at the Peterson Institute puts it this way: "The evidence on how AI is affecting the labor market today is inconclusive."
For Norwegian SMEs, this means that OpenAI's figures are an interesting indication, not a definitive answer. Use them to understand the direction of travel, not to calculate how many positions will disappear.
What should you do now?
The report points to tasks that cross professional boundaries as relevant areas to map. Here is a concrete approach:
-
Map the tasks. Write down the tasks you and your team perform that fall outside your own field. The accountant who creates landing pages. The marketer who writes contracts. The leader who conducts legal research.
-
Choose one area to test. Start with the task that is most repetitive and carries the lowest risk. Relevant examples include marketing copy, meeting minutes, or customer communication.
-
Set a measurable target. Define how much time the task takes now and what you want to use the freed-up time for. Without measurement, it will simply become another pilot project without a conclusion.
-
Build something simple and measure the result. The first AI agent can take over part of one generalist task and give the team experience with this way of working.
This is the core of how an AI Audit works: mapping, prioritization, and a concrete plan for which tasks should be automated, which require human quality assurance, and which should be left alone. Once the map is ready, you can take the next step with AI Kickstart to build the first automation. AI training teaches the team how to use and monitor the workflow. For a concrete five-step plan for testing an AI agent in an SME, see How to run an AI agent pilot in your SME: 5 steps.
A brief caveat about the method
The sample in the OpenAI study consists of US business users of ChatGPT. It does not consist of Norwegian SMEs, private individuals, or a global sample. The reported patterns, with customer experience and design at the top, resemble what we see in Norwegian companies, but they should be generalized with caution.
The report provides an opportunity to examine how tasks are distributed in your own company. The generalist role has long rested with the SME leader. The company can now test where AI provides practical support, with measurement and professional quality assurance.
Frequently asked questions about the OpenAI report
What is task crossover in the OpenAI report?
Task crossover means that a work task in ChatGPT is classified under an occupation different from the user's role. An HR employee asking for help with design or a designer asking legal questions is an example of crossover.
Why is this relevant to Norwegian SMEs?
Norwegian SMEs with 7-100 employees are already driven by generalists. One person handles accounting, marketing, HR, and customer follow-up. OpenAI's figures confirm that this pattern is common, including in larger organizations.
Does this mean AI is replacing people?
No. The study measures what people do with AI, not whether jobs are disappearing. OpenAI itself points out that the report says nothing about hiring, wages, or productivity. It describes behavior, not impact.
Is the report impartial?
OpenAI is both the publisher and a commercial company. The report is not peer-reviewed, and the classifier is OpenAI's own. External economists at the Peterson Institute say the evidence on AI and the labor market remains incomplete.
What should a Norwegian SME leader do with this information?
Start by mapping which tasks actually cross professional boundaries in your company. Then choose one area, such as marketing copy or contract review, and test an AI agent in a limited pilot project. Measure the time spent before and after, and require human quality assurance for high-risk tasks.
Want to know where AI can make a real difference in your company? Start with an AI Audit or contact us for a no-obligation conversation.
