What Happens to Professional Identity When AI Does the Work You Were Known For?

I had the opportunity to join a panel discussion during a recent Perceptyx town hall. Our CEO asked a few questions about how we use AI in our daily work. Then he turned to me. "Did you have any reservations when you first started making AI part of your daily routine?" I blurted out, "Did I have reservations? I think I had an AI-dentity crisis."

Then I told my entire company the rest. I still have an emotional reaction when AI produces in seconds something that used to take me hours to develop and refine. Whatever imposter syndrome had been hiding in me, AI found it and dragged it into the open. I started asking myself whether I loved AI or hated it, whether my own expertise still mattered, and if it did not, what unique value I brought now.

That last question decides whether AI adoption works inside an organization. The barrier to adoption is rarely the tool so much as the person holding it, trying to figure out who they are now.

Why Does Using AI Feel Like Losing Something?

A prime example came when I received accolades for a clever analogy I used to describe data in an executive presentation. Words of affirmation fill my bucket, yet I was not the least bit proud. I had not come up with the analogy. I gave ChatGPT a couple of prompts, and it produced the analogy along with talking points. Show off. The situation left me feeling like a fraud and slightly jealous of a piece of software.

Behavioral science has a name for part of what I was feeling. The IKEA effect, documented by Michael Norton, Daniel Mochon, and Dan Ariely in 2012, describes how people place higher value on things they helped build. Assemble the bookshelf yourself and you like it more than the identical one delivered finished. My analogy arrived finished. I had contributed the prompt and the judgment to use it, and neither felt like assembly.

The deeper issue is what organizational psychologists call professional identity: the bundle of skills, expertise, and contribution that defines who we are at work. Adapting to AI asks for new skills and new processes. It also asks for identity work, meaning we reinterpret ourselves while trying to hold onto confidence, belonging, agency, and purpose. Nobody puts identity work on the rollout plan, and it happens anyway.

How Common Is This Experience Across the Workforce?

As I worked through my own version of this, it became clear I was not alone. A peer told me he used to love meeting people smarter than him because it kept him in a growth mindset. He is now wondering how he grows with AI in the mix. Another peer told me she does not know whether it is good or bad to admit she uses AI. "Is this skill really a skill? Is it valued or looked down on? I used to be valued for being able to write in ways others couldn't. Who am I, if I am not the creative one?"

Perceptyx's Beyond the Hype study of more than 3,600 employees across North America and Europe found 8 in 10 employees have had to build new skills because of generative AI. Tasks have shifted for 78%. Fourteen percent feel less secure in their jobs. Confidence is wide but shallow, with 82% feeling at least moderately confident working in an AI-augmented environment while only half feel very or extremely confident. Thirty-eight percent are unclear on how AI will affect their roles.

Our Managing in the Age of AI research, drawn from 2,992 employees in July 2026, tells us where employees go with those questions. They go to their manager, and fewer than half (44%) say their manager helps them adapt to AI-related changes. Where that support exists, however, we found that 82% trust their organization to use AI responsibly and 81% intend to stay.

Should Employees Admit They Use AI?

My peer's question about whether to admit AI use is an impression management problem, and it cuts both ways. Does using AI make me look technologically capable, or does it make me look like I needed assistance? Every professional has curated an identity over years of work. Disclosing AI use feels like handing someone a reason to revise it.

A customer meeting made the stakes concrete for me. A seasoned recruiter was brought in to help shape candidate survey content. Midway through, she said that sometimes she feels like she "doesn't know how to recruit anymore." She has done this for 20 years. Now she assumes every resume was written by ChatGPT. She cannot differentiate candidates the way she used to, and she does not know whether visible AI use should count against a candidate or for them. She was questioning whether recruiting was still a strength of hers, and she was also, without saying so, questioning whether the candidates still had strengths she could see.

My colleagues wrote recently about how audiences now run a detection pass before an evaluation pass on professional work, and how that gives employees a rational reason to limit AI use or hide it. In our Generative AI research, 59% of individual contributors feel comfortable raising concerns about AI, compared with 72% of managers and 84% of executives. The people closest to the work feel least safe talking about how it is changing.

Where Does Human Judgment Still Matter?

PYX-Voice, the first benchmark from PYX Labs, tested seven frontier models on 84 employee listening tasks. When answers were clear and verifiable, the models passed 76% to 82% of tasks. When the work required weighing incomplete, emotional, or context-dependent signals and resolving them into one takeaway, scores dropped sharply. Synthesis was the lowest-scoring capability across every model, ranging from 14% to 57%.

Megan Steckler on our team pointed out that AI struggles with the same tasks humans do. Reading a room, holding two contradictory pieces of feedback at once, and deciding what a comment means for a specific team are hard for people as well as models. The difference is that experienced people have spent careers getting better at them. The 20-year recruiter cannot tell from a resume who wrote it. However, she can still tell, while sitting across a table or on a Zoom interview, who understands the job.

An Uber driver showed me what that looks like in practice. Last December I rode in a driverless car for the first time, felt every emotion available, and got car sick after two sudden stops. The next morning I took a human-driven car to the airport and asked my driver what he thought of autonomous vehicles. "At first, I did think I would be out of a job," he said, "but then I got curious. I wondered how these cars knew all of the ins and outs of San Francisco that took me years of experience to nail down." That curiosity led him to a job as an autonomous specialist for the company. He now uses simulation and closed-course testing to teach the cars a city he knows like the back of his hand. His expertise found a new use instead of disappearing.

What Can Leaders Do About an Identity Transition They Cannot See?

My own experiences, and the ones people keep sharing with me, have taught me that I am not afraid of the technology. What I need is time to be curious and to find my strengths inside it. That means acknowledging the discomfort. It means giving myself permission to redefine where I create value. That consists of the judgment and contextual knowledge I still have that the tool does not.

Organizations have to make the same move at scale but cannot do it blindly. Perceptyx customers have been asking for AI-specific survey content often enough that Bradley Wilson, our Global Head of Research and Insights, developed a set of AI items now available in our listening library. The items ask whether people feel confident adapting to AI-related changes, whether their manager supports experimenting with AI tools, and whether training is adequate. They belong in any engagement or pulse survey running in 2026, because they measure the identity transition directly.

Our Activating AI Adoption workshop runs that measurement through the COM-B model, which holds that behavior requires capability, opportunity, and motivation together. An AI adoption pulse survey establishes the baseline. A facilitated session can then help leaders read the results and design environmental nudges. Six AI adoption behaviors, now part of the People Insights Model, give Activate something specific to reinforce afterward, through Teams, Slack, and the tools people already use.

Nudges aimed at managers can support professional identity transitions like this, since managers are there to help employees mediate their encounters with AI. A nudge can prompt a manager to normalize mixed emotions about AI in a team meeting. It can prompt recognition for an experiment that did not pan out. It can suggest asking a direct report what part of their work they are proudest of and how AI changes it, or talking openly about the manager's own AI use so disclosure stops feeling like a confession.

Employees adapt faster when they believe AI changes how they contribute, not whether they matter, and the manager is the person who can say so credibly.

As organizations work out where AI performs well, they must also work out how their employees are experiencing the transition. Successful AI adoption depends on both, even though a deployment dashboard can report only the first.

Ready to Measure the Human Side of Your AI Rollout?

Schedule a meeting with our team to learn how an AI adoption pulse survey, the Activating AI Adoption workshop, and targeted nudges can show you where employees stand and help managers lead them through it.

For the research behind how employees across regions, generations, and job levels are experiencing AI, read Beyond the Hype: Global Employee Perspectives on Generative AI. For data on what employees now need from their managers, read Managing in the Age of AI.