Artificial intelligence is beginning to change the structure of work before we have a clear picture of what work will look like on the other side.
Much of the debate around AI and employment has focused on the destination: which professions will grow, which work will decline, and what new forms of work may emerge. The more immediate question is already visible inside roles that still exist.
A role can keep the same title while research takes a fraction of the time it once did. A first draft can arrive before a junior employee has learned how to build one. Routine cases can move to an AI system while the remaining human work becomes more complex. A professional who spent years becoming highly capable at one method of working may find that the method changes faster than the profession itself.
The new work that follows will not necessarily appear at the same time, in the same organization, or in a form that people can immediately reach.
That period between today’s changing work and viable next work is becoming one of the central workforce questions of the AI economy.
I call it the Transition Gap.
The Transition Gap is the period between work becoming materially different, less available, or less valuable and a person reaching viable next work.
The gap can begin long before a role changes formally, while a person is still employed and the organization is still deciding what the change means.
The World Economic Forum’s 2025 research, based on input from more than 1,000 employers representing over 14 million workers, finds that 39% of key skills are expected to change by 2030. That scale of change makes the period between familiar work and viable next work a leadership issue, not simply an individual career issue.
Work can change before the role does
Most jobs contain many different activities. AI does not need to replace an entire occupation to change the experience of working in it.
It can take over preparation work, produce a first analysis, handle routine requests, summarize information, or draft documents that previously required hours of human effort.
Some of these changes will make jobs better. People may spend less time on repetitive work and more time dealing with difficult problems, customers, exceptions, and decisions.
The transition still has consequences while the person remains employed.
The organization may begin expecting different capabilities from someone whose title has not changed. Review becomes more important than production. Framing the problem becomes more important when collecting the information becomes easier. Understanding the context around an answer becomes more valuable when producing the answer takes less time.
A person can therefore be inside a career transition without having changed jobs.
Organizations that wait for formal role changes before thinking seriously about workforce transition will see some of this movement too late. Leaders need to look at which activities are shrinking, what is becoming more important, and where the organization is beginning to rely on people differently.
That provides an early indication of where the transition has already started.
There is another consequence. Some of the work AI removes is also the work through which people learn a profession. When junior tasks move to AI, leaders need to ask what people were learning by doing that work and where that learning will now happen.
Reaching the next work is often the difficult part
New roles may absorb part of the change as existing work changes. Their existence does not mean the people affected can immediately reach them.
The next work may require experience the person has never had. It may sit in another part of the organization. A receiving manager may prefer someone whose previous title already resembles the new role. The opportunity can exist while remaining difficult for the person affected to reach.
This is one reason the workforce conversation cannot be reduced to eventual projections about the number of roles that may be created or changed.
A positive outcome for an organization or an economy can still contain a difficult transition for an individual.
The quality of that transition depends partly on how much time people have before their choices narrow.
Someone who is still employed has access to work, colleagues, projects, managers, and income. They can spend time with another team or try a new responsibility before making a permanent move. They can learn in the environment where the capability will actually be used. A receiving manager can observe how they work instead of making a decision from a résumé alone.
Those options become harder to create once the old work has already changed beyond recognition.
Workforce transition needs to begin earlier than many of our existing systems assume.
Training is part of the response
Training will be an important part of the workforce response. People will need to learn new tools, develop different capabilities, and understand how their professions are changing.
But a course does not create a career transition.
A person can complete training without ever being given the opportunity to use what they learned. They can earn a credential without gaining the practical experience a manager wants to see. An organization can train hundreds of employees for work that has only a small number of real openings.
The connection between learning and work has to become much tighter.
If an organization is preparing people for a new responsibility, leaders should know what that work looks like, where it exists, and how someone will gain their first meaningful experience doing it.
Sometimes the answer will be a rotation. Sometimes it will be supervised project work. Sometimes the existing role can change gradually enough for the person to develop new capability without making one large jump.
The important measure is movement into viable work, not simply participation in learning.
Organizations need workforce intelligence. People need career intelligence.
Companies are developing increasingly detailed views of their workforces. They want to understand which capabilities they need, where shortages are forming, and how AI may change demand.
The individual has a different problem.
A professional needs to understand how their own work is changing, what their experience has produced, and where that experience may become useful next.
I think of this as Career Intelligence.
It begins with a better record of the work itself. Projects, difficult decisions, repeated problems, and lessons provide a richer picture than a job title alone.
AI may eventually help people interpret that record by surfacing recurring patterns or comparing experience with changing work. But the system should support the person rather than make the career decision for them.
Organizations can provide better information and better opportunities. The career still belongs to the individual.
The leadership work has already started
No leadership team knows exactly what the workforce will look like five or ten years from now. They do not need to.
Organizations can already see which work is changing. They can begin career conversations earlier, connect learning to real work, make internal opportunities easier to see, and examine what happens to professional development when tasks move to AI.
They can plan the movement of people as deliberately as they plan the deployment of technology, especially where work is changing in one part of the business and new capability is needed in another.
They can also give people opportunities to test what might come next while they still have access to the work they know today.
For the person experiencing the transition, the change may begin quietly.
A task that once took an afternoon takes twenty minutes. A first draft is already waiting.
Routine work stops arriving.
The job is still there, but the work through which the person built a career has started to move.
