Artificial intelligence is changing more than the speed of work. It is also changing who performs specific tasks inside organizations.
New research from OpenAI Economic Research found that 16.8% of work-related ChatGPT messages and 43.5% of occupation-specific messages concern tasks associated with another occupation. The findings suggest that workers are using AI to cross traditional job boundaries before companies formally update job descriptions, workflows or titles.
A small-business owner can draft marketing copy, review a contract or complete basic financial analysis without immediately turning to a specialist. A salesperson can explore customer data that might previously have gone to an analyst. A marketer can troubleshoot a website without waiting for an engineer.
OpenAI describes this pattern as “task crossover”: work historically associated with one occupation appearing in the AI use of people employed in another. The research indicates that AI may reorganize many jobs by allowing workers to absorb tasks that were previously handed off to other functions.
Nearly Half of Occupation-Specific AI Use Crosses Job Boundaries
The study analyzed more than 800,000 messages from US ChatGPT users.
Researchers first separated generic work from occupation-specific activity. Generic tasks included writing, summarizing and scheduling because these activities appear across many jobs and therefore do not provide strong evidence of occupational crossover.
Of the messages classified as non-generic, 43.5% involved tasks outside or beyond the user’s own occupation. The remaining 56.5% stayed within or close to the user’s field.
This result matters because it suggests that AI use is not limited to helping people perform familiar work more efficiently. A significant share of users are also applying it to activities normally assigned to someone with a different professional role.
These patterns may appear in usage data before employers recognize them through new organizational structures or revised job descriptions.
Customer Experience Workers Show the Highest Crossover
Task crossover varies substantially by occupation.
After generic activities were excluded, outside-occupation tasks represented 77% of occupation-specific messages from customer experience workers. This was the highest share among the occupational groups examined.
Customer experience roles often sit close to operational problems. Workers may need to understand product issues, create communications, analyze customer behavior or troubleshoot technical concerns before they can resolve a case.
AI can reduce the need to transfer every issue to another department. A customer experience employee may draft technical explanations, examine data or prepare marketing-style communications directly.
This does not necessarily mean that specialist functions disappear. It suggests that front-line workers can address a broader share of problems before escalation becomes necessary.
Designers and Human Resources Workers Also Expand Their Roles
Designers recorded an outside-occupation share of 75%, while human resources workers reached 69%.
Designers appear to draw heavily from other fields. Around 35.2% of all messages from designers involved work normally associated with another occupation. Yet design tasks represented only 1.7% of messages from workers in other fields.
This means design is primarily an importer of tasks rather than a major exporter. Designers use AI to take on activities from marketing, business, technology or other domains, while workers outside design are less likely to perform traditional design work.
Human resources professionals show a similar pattern of role expansion. Their work can involve policy interpretation, employee communication, data review, recruiting content and operational problem-solving. AI may allow them to handle more of these adjacent tasks without immediately relying on legal, finance or analytics teams.
Legal and Marketing Roles Also Cross Traditional Boundaries
Outside-occupation work accounted for 56% of occupation-specific messages from legal workers and 53% from marketers.
Legal professionals may use AI for financial calculations, operational research, technology troubleshooting or communication tasks that fall outside conventional legal work.
Marketers appear to operate in both directions. They use AI for tasks associated with other professions, while marketing work also spreads widely across the organization.
About 24.3% of marketers’ messages concerned tasks from other occupations. At the same time, marketing tasks represented 8.9% of messages among workers in other fields, the highest outward share in the sample.
This suggests that marketing is becoming both more interdisciplinary and more distributed. Marketers absorb analytical, technical and operational work, while employees in sales, design and other departments create promotional materials themselves.
Engineering Tasks Travel Across the Organization
Engineering displays a different pattern.
Only 18.5% of engineering messages involved tasks from other fields, indicating that engineers remain relatively focused on work close to their core occupation. However, engineering tasks accounted for 7.4% of messages among workers in other professions.
This makes engineering an important exporter of work.
Employees outside engineering are increasingly using AI to troubleshoot software, interact with technical systems or resolve basic digital problems. A task that once required a developer or IT specialist may now be completed by the worker who first encounters it.
This shift can reduce waiting times and internal handoffs. It may also allow engineers to concentrate on more complex technical work.
The risk is that workers may attempt tasks beyond their expertise without recognizing security, quality or reliability limits. Organizations may therefore need stronger rules defining which technical activities can be completed independently and which still require specialist review.
Financial Calculations and Troubleshooting Spread Widely
Some tasks cross occupational boundaries more often than others.
Financial calculation and technology troubleshooting ranked among the three most common outside tasks in all seven other occupational groups examined.
This means workers across design, marketing, sales, legal, customer experience and human resources are using AI to perform activities historically associated with finance or engineering.
Basic financial analysis can help employees evaluate budgets, compare scenarios or understand performance without waiting for a finance team. Technical troubleshooting can help them resolve website, software or system issues more quickly.
These tasks are particularly suitable for crossover because they often appear as immediate operational needs. The worker closest to the problem may want a fast answer rather than initiating a formal handoff.
However, widespread access does not eliminate the need for verification. Errors in financial calculations or technical advice can still create material consequences.
Marketing Work Travels Far Beyond Marketing Teams
Creating marketing materials appeared across five other occupational groups and was especially common among design users.
AI lowers the cost of producing first drafts of advertisements, product descriptions, customer emails, social posts and campaign ideas. This makes marketing work accessible to employees who may not have formal training in the field.
A salesperson may create outreach material. A product employee may draft a launch message. A small-business owner may prepare a campaign without an internal marketing department.
This can increase speed and experimentation. It can also fragment brand control if many employees create public-facing material independently.
Organizations may need shared templates, review processes and brand guidelines to benefit from distributed marketing without sacrificing consistency.
Smaller Businesses Show More Task Crossover
The research found that business size affects how workers use AI.
Among average users, outside-occupation work represented 18.9% of activity in workspaces with two to five seats. The figure declined to 16.3% in workspaces with more than 100 seats.
Smaller organizations typically have fewer specialist teams and less formal division of labor. The person who encounters a problem is more likely to solve it directly rather than pass it to another department.
AI can serve as a generalist tool in this environment. It allows a small team to access capabilities associated with marketing, finance, legal review, research or technical support.
Larger companies generally have more established functions and internal resources. Workers may therefore continue to use traditional handoff processes even when AI could support the task.
The pattern was less consistent among the heaviest AI users. OpenAI suggests that advanced users may develop stable workflows that look similar across organizations, regardless of company size.
AI May Reduce Internal Handoffs
The research points to a changing division of labor inside organizations.
Many business processes involve handoffs. A salesperson sends data to an analyst. A marketer asks a developer to fix a website problem. A small-business owner sends a contract to a lawyer for an initial review.
AI may allow the first person involved to complete part or all of the task independently.
Reducing handoffs can shorten project timelines, lower coordination costs and improve responsiveness. It may be particularly valuable for small businesses that cannot employ specialists in every area.
Yet the effect will differ by task. Some activities can be safely handled with AI assistance, while others require professional judgment, accountability or formal approval.
The likely outcome is not the complete elimination of specialist work. Instead, specialists may receive fewer basic requests and spend more time on difficult cases.
Jobs May Reorganize Before Titles Change
Traditional labor-market research often begins with a fixed list of tasks assigned to each occupation. It then asks which tasks AI can perform.
OpenAI’s findings suggest that this approach may miss an important development: the task list itself is changing.
Workers are experimenting with new combinations of activities before firms update their organizational structures. A job title may remain the same even as the daily work becomes broader.
This can make conventional employment data slow to reflect the effects of AI. Occupational statistics may show stable job categories while the actual content of those jobs changes significantly.
Usage data can therefore serve as an early indicator of labor-market transformation.
Task Crossover Could Increase Worker Autonomy
AI-supported crossover may give workers more control over their work.
Employees can solve problems directly, test ideas and complete tasks without waiting for another team. This can increase autonomy and make roles more varied.
A salesperson who can analyze customer data may make faster decisions. A designer who can evaluate campaign performance may connect creative work more closely with business outcomes. An HR employee who can draft operational analysis may respond more quickly to workforce issues.
Greater autonomy can also increase responsibility. Workers need to understand when an AI-generated answer is sufficient and when specialist validation remains necessary.
Training will therefore need to cover not only how to use AI, but how to judge its output and recognize professional boundaries.
Specialists May Shift Toward Higher-Value Work
Task crossover does not necessarily imply that specialists become less important.
If generalist workers handle routine calculations, troubleshooting or first drafts, specialists may focus on more complex analysis and higher-risk decisions.
Finance teams could spend less time preparing simple comparisons and more time on capital planning or control. Engineers could receive fewer basic support requests and concentrate on architecture, security or performance. Legal teams could focus on negotiation and risk rather than preliminary summaries.
This shift may improve productivity, but only if organizations redesign workflows effectively. Without clear processes, specialists may still need to correct low-quality work after it has already created confusion.
The productivity benefit depends on distributing tasks intelligently rather than merely expanding every employee’s workload.
New Skills Will Become More Valuable
As workers take on tasks outside their original occupations, several capabilities become more important.
The first is problem framing. Users must explain what they need clearly enough for AI to provide useful support.
The second is verification. Workers must review calculations, technical guidance and written output rather than accepting results automatically.
The third is cross-functional literacy. Employees do not need to become full experts in every field, but they need enough understanding to recognize poor advice and know when to escalate.
The fourth is judgment. AI can support execution, but responsibility for consequences remains with people and organizations.
These skills may become central to future workforce development.
Risks of Expanding Work Beyond Formal Expertise
Task crossover creates operational benefits, but it also introduces risks.
A worker reviewing a contract with AI may miss a legal issue. An employee performing financial analysis may use the wrong assumptions. A marketer troubleshooting a website may introduce a technical or security problem.
The easier it becomes to attempt unfamiliar work, the more important governance becomes.
Organizations may need to classify tasks by risk. Low-risk drafting or exploration can be completed independently, while high-impact decisions require review by qualified specialists.
Data protection is another concern. Workers using AI for legal, customer, financial or technical tasks may handle sensitive information.
The goal should be controlled expansion of capability, not unrestricted substitution for expertise.
The first issue is where crossover already occurs. Employers should examine which teams are using AI for tasks outside their formal roles.
The second is whether those activities create value. Faster completion matters only if quality remains acceptable.
The third is where specialist review is still required. Companies should define clear escalation rules for legal, financial, security and other high-risk work.
The fourth is training. Employees need domain awareness, verification skills and guidance on responsible AI use.
Finally, firms should monitor whether AI is reducing handoffs or simply adding more work to existing roles. Role expansion should improve outcomes, not create hidden overload.
OpenAI’s research suggests that AI is already changing the boundaries of work.
Among occupation-specific ChatGPT messages, 43.5% involved tasks associated with another profession. Customer experience, design, human resources, legal and marketing workers showed particularly high levels of crossover.
Engineering and marketing tasks were among those that traveled most widely across occupations, while financial calculations and technology troubleshooting appeared throughout the workforce.
The findings indicate that AI is not merely automating existing task lists. It is allowing workers to perform new combinations of activities before organizations formally redesign jobs.
AI is turning many employees into broader generalists while allowing specialists to focus on more complex work. The central management challenge will be deciding which tasks can safely cross job boundaries and which still require formal expertise, review and accountability.
