Essays · AI and work · Leadership · Psychology

Leading people through AI without breaking their sense of worth

How you roll out AI teaches people what you think they are worth. A leader's guide to telling the truth, protecting identity, and making room for grief.

Picture someone who has spent twenty years getting very good at one kind of work. They hand a task they have done a thousand times to an AI tool, and it hands back something solid in seconds.

The first feeling is relief. The second, a beat later, is a small, cold question: if it can do that, what am I for?

That is not a technology question. It is an identity question. And if you lead people right now, how you help them answer it will be remembered long after they have forgotten the name of the software.

I come at this from three sides: as a school psychologist who has watched how people react to unexpected news about themselves, as a principal and superintendent who has announced changes people did not ask for, and as a former senior manager at Oracle who now builds AI-powered platforms for education.

The map: tasks, roles, self

In Who Am I When AI Can Do My Job?, I describe working identity in three layers.

Tasks are what you do. Draft the memo. Close the books. Answer the ticket. Tasks are concrete and measurable, and that is exactly why AI reaches them first.

Roles are who you are at work. The expert. The fixer. The mentor. The calm one in a crisis. AI reshapes roles, sometimes a lot, but rarely erases them, because roles live in relationships.

Self is who you are. Your values, your character, your relationships, your commitments. AI cannot touch your self. The only way to lose it is to hand it over.

The pain comes from what I call fusion, when a person’s self gets welded to their tasks. After two decades of being the one who does a certain thing best, it is easy to forget where the work ends and you begin. Then a machine does the work, and it feels like it wrote you out of the story.

A leader’s first job is to help people pull those layers apart.

The rollout is the message

Every leader thinks an AI rollout is about technology, budgets, and workflows. It is. But your people are listening for something else. They want to know one thing: do I matter here?

The memo answers that question. So does the timeline, who gets told first, and whether anyone asked the people doing the work what they think. Whether the savings get announced before the plan for the people answers it loudest of all.

You can get the technology right and still get this wrong. If you do, you will pay for it in trust, which is the most expensive thing to rebuild.

Tell the three-part truth

The single biggest mistake leaders make in an AI rollout is trying to manage feelings instead of telling the truth. Leaders soften and spin. They say “no one will lose their job” when they do not know that. They say “this is exciting” to people who are scared.

There is a better approach. I call it the three-part truth:

  • What we know. What the tool is, when it starts, and what it will and will not do.
  • What we don’t know yet. How it changes each role, and how well it really works on ordinary days.
  • When we’ll know more, and how you’ll be involved. A date, a rhythm of updates, and a real role for the team in figuring it out.

People can handle uncertainty. What they cannot handle is being lied to or left in silence. And if jobs really are going away, say so early. The kindest version of hard news is the honest one, delivered soon, with real support attached.

Do not spend savings you have not seen

Be careful about banking AI savings early. In McKinsey’s 2026 survey of business leaders, about a third of companies had expected the year before that AI would shrink their workforce, but only 14 percent said it actually had.

In 2025, Australia’s Commonwealth Bank reversed 45 customer-service job cuts it had linked to a new AI voice bot and apologized to staff. According to the workers’ union, call volumes went up, not down.

Prove the tool works on real Tuesdays, not in the demo. Learn where it breaks. Then make decisions about roles with the people who do the work at the table.

Let the people redesign the work

Before you finalize anything, sit down with each team and ask four questions:

  1. What do you do that the tool could handle?
  2. What do you do that the tool cannot, or should not?
  3. Where will the people we serve need a human, no matter what?
  4. If you got back ten hours a week, what would you do with them that would make this place better?

Your frontline people know where the tool will fail and which “routine” requests are not routine at all. Asking them makes the rollout work better. It also tells them something powerful: we are doing this with you, not to you.

And resist turning new tools into a way of watching people. When people feel watched, they stop admitting mistakes, including the tool’s mistakes. Amy Edmondson’s research on psychological safety points the same way: teams that feel safe enough to report problems are the ones that catch and fix them.

Pay for the bridge, and protect the bench

If you are asking people to change how they work, you have to pay for the bridge they walk across. That means real training during work hours, with practice on real tasks and someone to ask when it gets confusing. As of 2026, only about a quarter of American workers told Gallup their organization had communicated a clear plan for AI.

It also means protecting your future bench. In August 2026, a Stanford team led by Erik Brynjolfsson reported that employment of 22- to 25-year-olds in the jobs most exposed to AI had fallen to about 19 percent below where it would have been, mostly because companies were hiring fewer young people. The researchers call these early signals, not proof. But if you stop growing junior people because AI handles junior tasks, ask where your senior people will come from in ten years. Expertise has to be grown.

Make room for the grief

A lot of what looks like resistance is grief. People are losing a version of their work they were proud of.

Name it once, out loud: some of you spent years getting great at work that is changing fast, and if this feels like a loss, that makes sense to me. Honor the craft before you change it, and thank people by name for what they built. Ask what they are afraid will be lost, and take the answers seriously.

And do not rush acceptance. Push people to be “on board” in a week and you will get agreement on the surface and résumés underneath.

You were never your job description

The deepest message a leader can send in this moment is that a person’s worth is not their output. The self your people bring to work, their care, their values, their relationships, was never the machine’s to take. Lead like you believe that, and they will too.

The next play

  • Draft your next AI announcement in three parts: what we know, what we don’t know yet, and when we’ll know more.
  • Before you finalize any role change, ask each affected team the four redesign questions.
  • At your next all-hands, name the loss once, out loud.
  • Thank one person by name this week for the craft they built.
  • Keep at least one entry-level apprenticeship path in your plan, even if it costs you some efficiency this quarter.