Most of us have accepted that AI is changing our work. Far fewer of us have changed how we see ourselves at work.
By Geralynn Madonna
Almost every conversation about work right now comes back to AI.
Will it replace jobs? Which ones? How soon? In a December 2025 Resume Now survey of 1,006 U.S. adults, 51% said they were worried about losing their jobs to AI in 2026, and 67% said they think AI will eventually threaten their job.
Those are reasonable concerns. But I think they leave out something important.
Many people have already changed. They have learned new tools, taken on new responsibilities, and quietly rebuilt the way they do their work. What has not changed, for many of them, is who they believe they are at work.
Think about a marketing manager who has spent the past year learning to work with AI.
At first, she experimented because she had to. Then she began building the workflows her team now relies on. She redesigned how campaigns are researched, drafted, and tested. Colleagues from other departments come to her with questions.
When her company posts a new role leading AI adoption across the business, she reads the description and recognizes every task on it.
She doesn’t apply.
“That’s for someone more technical. I just figured out what we needed.”
The new version of you may already be doing the work
Ask her whether she can do what that job requires, and she would probably say yes. She has done most of it already. The evidence is not in question.
Ask her whether she is someone who leads AI work, and she hesitates.
She still introduces herself by her old title. When she describes what she does, the new work comes last, almost as an aside: something I picked up along the way. Her sense of who she is at work remains tied to the role she held before any of this began.
She is not unusual.
Across many organizations, people are already doing work that did not exist in their job descriptions two years ago. An accountant has become the person who checks what the AI got wrong. A teacher designs lessons around tools her students use every day. A customer service lead now trains the system that answers the first round of questions.
They have adapted. In many cases, they have adapted well.
But adapting to a change and recognizing yourself in it are not the same thing.
The capable version of these people is already there. What is missing is not ability, effort, or even confidence. What is missing is recognition: the point at which they see that capable version and think, that is me.
This is where Recognition Lag differs from what is often called imposter syndrome. Imposter syndrome usually describes someone who doubts they are as capable as others believe. The marketing manager does not doubt it. She knows she is good at this work.
What she has not done is recognize that work as a description of who she is. A person can know they are competent and still not identify with the role that competence points toward.
When we accept the change but don’t yet recognize ourselves in it
I call this experience Recognition Lag: the period in which we have accepted something as true about ourselves, but have not yet recognized it as part of who we are. It is one of the concepts within Recognition Theory, which I developed to explore how change becomes part of our sense of identity.
Recognition Lag is not denial. The marketing manager knows exactly what she has done. It is not simply a lack of confidence, either. She can be confident in her work and still not recognize herself in the role that work points toward. The two can exist at the same time.
One of the clearest signs of Recognition Lag is the word but.
I can build these systems, but I’m not really a technical person.
I’ve been leading this change, but I’m not a leader.
I know how to do this now, but that’s not what I do.
During the lag, we tend to keep new evidence contained. We limit it to one setting: I can do this on this project. We credit it to circumstances: anyone could have done it with these tools. Or we treat it as temporary: it’s just until things settle down.
AI adds a containment of its own. It is easy to give the tool the credit.
The AI did most of it. I just told it what to do.
But knowing what to ask, recognizing what is wrong with the answer, and deciding what the work needs are exactly the skills that make someone valuable now. When we credit the tool, we keep accepting the evidence while protecting the earlier version of ourselves.
This may be one of the most significant ways AI affects how we see ourselves. In the past, when we learned to do something new, the evidence was clearly ours. Now there is always another explanation within reach. Whatever we accomplish, we can attribute it to the tool.
And the more capable the tools become, the more reasonable that explanation sounds. Every success arrives with a ready-made reason it doesn’t count.
Over time, a person can become highly skilled at work they still describe as the tool’s. The capability grows. The recognition does not.
There is another reason the lag may be especially common right now. Identity usually changes slowly. We need time for new experiences to feel representative of who we are. Work, however, is changing faster than many of us have ever experienced. The distance between what we do and who we believe we are can grow quickly, and it can grow without our noticing.
The cost of not recognizing yourself
Recognition Lag can be brief and harmless. Sometimes it is wise to wait before deciding that a new experience reflects who we are.
But when the lag lasts, it can carry a cost. I call that cost Recognition Debt: the accumulated consequences of not recognizing who we have already become.
Recognition Debt is rarely visible. No one sends us a notice for the job we didn’t apply for. There is no record of the raise we didn’t ask for, the project we let someone else lead, or the introduction in which we described ourselves by a role we left behind.
Yet those moments add up.
The marketing manager does not lose her job. Nothing dramatic happens. Someone else simply gets the role she was already doing. A year later, she is training that person.
The cost is especially high right now. In a Gallup survey of more than 22,000 U.S. workers in late 2025, only 28% said it was a good time to find a quality job, down from 70% in mid-2022. When opportunities are harder to find, the ones that do appear matter more. Passing one up because it doesn’t yet feel like me is not a small decision.
The debt can also follow us into conversations about our future. When people worry that AI will replace them, they are often picturing the job they used to do. They may be measuring their risk against an earlier version of themselves, rather than the person who has spent the past year learning, adapting, and solving new problems.
Organizations carry a version of this debt as well. Many companies are now training people for new kinds of work. If employees develop new capabilities but do not recognize themselves in new roles, much of that investment stays hidden inside job titles that no longer describe the work.
Why another course may not close the gap
The usual advice about AI is to keep learning. Take the course. Earn the certificate. Build the skills.
That advice matters. But for many people, skills are not what is missing.
The marketing manager does not need more evidence that she can do the work. She already has a year of it. Another certificate would give her one more thing she can do, and she might still believe it does not change who she is.
This is one of the more surprising things about recognition. Repeating an experience can strengthen what we believe we can do without changing whether we see it as part of who we are. We can collect proof for years and continue explaining it away.
What tends to make a difference is not simply more evidence, but evidence that becomes harder to contain.
When we succeed in more than one setting, it becomes harder to say the ability belongs to one project. When the results continue after the circumstances change, it becomes harder to credit the situation, or the tool. When we have been doing the work long enough, it becomes harder to call it temporary.
Sometimes the shift comes from rethinking what the identity itself means. If she believes that “technical” means writing code, she may never see herself that way. If she recognizes that the work now depends on judgment, design, and knowing what to ask, she may realize she has been doing technical work all along.
And sometimes a single moment matters more than months of practice. A crisis arrives, the system fails, and everyone turns to her. She solves the problem without hesitation, and in that moment, the evidence is no longer something she does. It is something she is.
Recognizing who you have already become
If the gap is about recognition, then the first step is noticing where it shows up.
Listen for the but. When you describe your work, do you add a qualifier to the parts that are new? I’ve been running this, but… The qualifier often marks the place where recognition has not yet caught up.
Notice how you introduce yourself. Does your description match the work you do today, or the work you did before things changed?
Read a job description the way you would read it for a colleague. If someone you respected had done exactly what you have done over the past year, would you tell them they were qualified? If the answer is yes, it may be worth asking why the same evidence does not count when it is yours.
And pay attention to what you credit to the tool. AI may have done part of the work. But someone decided what the work needed, recognized when the answer was wrong, and knew what to do next. That someone was you.
For managers, the question is similar. The people most ready for new roles may not be the ones asking for them. Some of your most capable employees may already be doing the work and still describing themselves by the job they had before.
AI will continue to change what our work looks like. Much of that is outside our control.
But how we see ourselves within that change is not entirely outside our control.
Perhaps the greater risk is not that AI will replace who we are.
It is that we will keep introducing ourselves as someone we no longer are, while the version of us that has already learned, adapted, and grown waits to be recognized.
Geralynn Madonna writes about identity, self-perception, and the psychology of personal change. Her work explores Recognition Theory and why becoming someone new doesn't always mean recognizing ourselves differently. Explore more at geralynnmadonna.com.
Sources
Resume Now. “60% of U.S. Workers Expect AI to Eliminate More Jobs Than It Creates in 2026.” 2026 AI & Job Security Outlook Report, January 12, 2026. Survey of 1,006 U.S. adults conducted December 9, 2025. https://www.resume-now.com/job-resources/careers/ai-job-security-outlook
Rugaber, Christopher. “It’s a bad time to hunt for new jobs, most US workers say in new Gallup poll.” Associated Press, March 24, 2026. Reports a Gallup survey of 22,368 U.S. workers conducted October 30 to November 13, 2025. https://www.opb.org/article/2026/03/24/job-hunt-market