The most unsettling question about artificial intelligence at work may not be whether your occupation disappears.

It may be whether the occupation keeps its name while quietly becoming a different occupation.

A teacher remains a teacher. A designer remains a designer. An administrator remains an administrator. A journalist remains a journalist.

But some of the activities through which those people learned to recognise themselves as competent begin moving elsewhere.

Research from the OECD offers a useful distinction. AI exposure does not mean that an occupation will necessarily disappear. Its 2026 occupational exposure measure was designed to examine how AI capabilities may affect occupations, skills and education over the coming five to ten years. Earlier OECD work found that many workers exposed to AI are unlikely to need specialised AI expertise themselves. Instead, AI may alter the tasks they perform and the skills their jobs require.

That distinction matters psychologically.

We often talk about work as if employment were a container.

You have a job or you lose it.

But identity does not work in containers.

Imagine someone who has spent fifteen years becoming extraordinarily good at preparing reports. Their judgement is partly embedded in researching, structuring, drafting, revising and explaining those reports.

Now imagine that software produces a credible first version in seconds.

The worker is still employed.

Their salary may be unchanged.

Their title remains on the door.

Yet something has happened.

A task that once provided evidence of competence has become something they supervise.

This does not automatically make their work worse. Automation can remove repetition, expand capability and create room for more valuable work. The mistake is assuming that because employment survived, nothing personally significant changed.

Work gives people more than income.

It gives repeated evidence.

I know how to do this.

People ask me for this.

I have become better at this.

This is part of what I contribute.

When technology changes those signals, workers may have to renegotiate identity before they ever negotiate unemployment.

That suggests a different question for the AI era.

Not only:

Which jobs will AI replace?

But:

Which parts of ourselves have we allowed particular tasks to certify?

If drafting disappears from your day, are you less of a writer?

If software diagnoses patterns faster, is expertise only pattern recognition?

If an algorithm produces twenty visual concepts before breakfast, where does the designer locate design?

Perhaps the most durable professional identity will not be built around ownership of a task.

It will be built around judgement, responsibility, curiosity, interpretation, relationships and the ability to decide what deserves to be done.

Technology may continue moving the boundary between human and machine activity.

Identity Sovereignty asks us to move another boundary too.

The boundary between what I currently do and what makes me valuable.

Those were never the same thing.

AI may simply make it harder to keep pretending they were.