How to Spot AI Identity Threat on Your Team This Quarter

Today’s hook

A head of operations told me last week that two of her senior people had quietly started doing less in their own roles. They were not failing reviews, not pushing back, not asking for help. They were just slightly elsewhere. The deliveries kept landing. The energy did not. She read it as motivation, ran a calm one-on-one with each, and got the same shape of answer twice: “I’m not sure what I bring anymore.”

What the research shows

A paper out this month in the European Journal of Investigation in Health, Psychology and Education studied 507 working professionals who use AI as a routine part of their day. The researchers ran a three-wave survey across three time points, then tested how the pieces hang together using structural equation modeling. The pattern is steady.

When working adults feel that AI is taking away the part of the job that used to make them feel skilled, or the part that used to feel autonomous, the felt loss becomes what the researchers call a professional identity threat. Professional identity threat then drives cyberloafing: the small drifts away from one’s own work that line managers usually read as low effort, distraction, or attitude. The identity layer sits between the AI moment and the drift, and the drift does not start without it.

The mitigator the paper finds is also specific. People who had internalized that working with AI is part of who they are professionally now (the paper calls this AI-inclusive identity) showed a weaker identity-threat pathway, especially on the autonomy side. The same prompts produced less threat, and less drift.

This sits inside a broader pattern catalogued in a peer-reviewed systematic review out in late March. When AI is appraised as a threat to identity-relevant resources, the behavior on the other side is withdrawal-shaped, not engagement-shaped. That dual-appraisal pattern is one of the consistent threads across the recent AI-and-work literature, and the framework piece on identity research every leader can read is now backed by a specific mediation finding rather than only a model.

What it doesn’t tell us yet

The paper is one study, one country (Chinese professionals), one set of survey items. The design measures the same people across three time points, which is stronger than a one-shot survey, but each piece is still self-reported, so the causal direction is theoretically motivated rather than experimentally proven. The outcome the paper measures is cyberloafing specifically, the small drifts to non-work, not broader disengagement, attrition, or burnout. The AI-inclusive identity construct is also at the early stage of its operationalization. Treat this as one well-designed study showing a real signal, not as a settled effect size for any one team. The pattern is consistent with the wider AI-and-work literature; the specific identity-threat-to-drift path is the new part.

One thing to notice in your work today

When someone on the team is visibly elsewhere around the AI-heavy parts of the week, the small browsing, the side-task spillover, the meeting where the energy is just slightly off, the research suggests it is worth asking, quietly, whether this is a discipline moment or an identity moment. The lever the paper points to is not monitoring or pushing harder. It is closer to what we have called trust not training in the rollout literature: naming the parts of the work that are still distinctly the person’s, redesigning the role with rather than around them, and being honest about what the tool is replacing. The shape of “what I bring” is being rewritten in the working life of the team this quarter. The work is helping the rewrite be visible, before reliance erodes self-belief in the people you most want to keep.