Yesterday, I was talking with a senior executive at a large multinational corporation. I have realized that every conversation I am having in the recent past invariably becomes about AI – the rollouts, the pressure, the pace of change and the impact it is having on all of us. I’ve been writing about it for some time now. At some point, the mood of the conversation shifted and in a rare, candid moment, he said something that stopped me.

“I worry about the impact AI is having on me, as a leader just as much as I worry about what it’s doing to my team.”

I’ve been sitting with that line ever since.

I have been thinking about how AI is affecting me. I have started noticing how it is affecting other leaders that I know and try and see if there has been any shift in their behavior or attitude in the past few years and realized that the majority of the leaders that I’ve known for years are also feeling the impact of AI, a handful of them having the time of their life.

While the majority of them struggling to either keep pace with the shifting technological landscape, and the impact all of this is having on their teams and their own ability to lead them through this liminal phase we are living through.

We have spent two years building frameworks for employee AI anxiety. We are learning how to communicate and lead change, how to reassure our team and how to bring people along the journey.

The entire conversation has been pointed outward, from us as leaders, to the people we lead. We have thought of leaders as the calm center, as the translators of uncertainty into direction.

But what about the leaders themselves?

The Speech That Doesn’t Land

Almost every leader I know of has given some version of the AI reassurance speech, as part of our team meetings, our all-hands, the town hall.

A speech where we laid out the plan for and the need to adapt with the changing times and adopt the phenomenal technological shifts of our times, named the opportunities that this could provide us, and ended with something like: “This is a tool like any other that we will use, not a replacement. This will help improve our productivity and that we’re going to navigate this together.”

And then watched it land with way less impact than we expected.

Here is why.

Research published in Harvard Business Review, identifies three deeper fears that AI actually activates in workers (Why Gen AI Feels So Threatening to Workers):

1. Competence: Am I still capable? If AI can do what I do, am I still skilled, still valuable, still the person I’ve built my professional identity around?
2. Autonomy: Do I still have meaningful control over my work? Or am I becoming a reviewer of outputs I didn’t create, approving things I don’t fully understand?
3. Relatedness: Do I still matter here? Am I still seen, still needed, still part of something that requires me specifically?

This is very similar to what Daniel Pink had mentioned in his best-selling book Drive – Autonomy, Mastery and Purpose. Every one of these is becomes a question mark and that creates anxiety and not drive.

The reassurances delivered by leaders don’t touch any of these and are trying to answer a question that employees aren’t really asking. That’s why the speech lands hollow.

And when you add the overall global decline in trust that leaders and institutions face, we are really in a bit of a situation that needs addressing.

But here’s the question that keeps pulling at me after that conversation I had with the executive of the large MNC.

These three fears:  autonomy, competence (mastery) and meaning (purpose) are also affecting the leaders as much if not more, as they not only have to take care of themselves but also take care of their teams and navigate this liminal state together, as one team.

The Mirror

Let’s think about what it means to be a senior leader right now.

You have built a career on the quality of your judgment, on synthesizing complexity, reading rooms and situations and making decisions that others couldn’t. And now there is a tool that can synthesize faster, research deeper, and produce polished outputs in seconds. You know you should be using it. You are using it. Yet something feels off, not in what it produces, but in what you feel while watching it produce.

Or think about decisions. You used to trust your gut informed by years of experience. Now there is an output on the screen, confident and comprehensive, and you are not always sure whether you are evaluating it or just ratifying it. Where exactly does your judgment begin? What about the times when your gut instinct is telling the exact opposite of what the AI model is telling you, confidently?

Amy Edmondson, whose work on psychological safety is among the most cited in organizational research, wrote a piece this week, documenting a pattern she had not seen before: employees are now using AI’s perceived neutrality to surface concerns they will not raise directly with their managers (What Leaders Need to Know About AI and Psychological Safety). The AI model had become the safe confessional.

This leads to questions like – does the team even need me? What value am I bringing to the team, when the AI models can do the research, make the decisions, recommend how to overcome personnel issues and sometimes even take actions?

And then there’s something harder to name, a discomfort around accountability. When AI is in the chain and something goes wrong, what am I actually responsible for? If I did approve the recommendation made by the AI model, does it equal to me having made the decision?

When I started researching about this, I realized that these are not just my personal reflections. There has been a lot of academic research (Cao et al., 2021; Suseno et al., 2021; Zhang et al., 2026) done in this area.

The research identifies these exact pressures as the core sources of AI anxiety among organizational leaders:

• threats to status and competence perception,
• erosion of decision-making autonomy, and
• deepening uncertainty around ethical accountability.

Leaders are experiencing the same competence, autonomy, and meaning threats they are trying to help their teams navigate.

What Suppressed Anxiety Does

A four-year ethnographic study, involving  more than 750 hours of direct observation and more than 300 interviews with leaders across multiple organisations found a clear and unsettling pattern: AI and digital transformations fail not because employees resist, but because leaders fail to manage their own anxiety (How Leadership Anxiety Derails Transformation, Fitzsimons et al., MIT Sloan Management Review, 2026).

When leaders suppress their anxiety, it doesn’t disappear. It disguises itself.

It becomes a vision that sounds bold in the all-hands and unconvincing in the corridor. It shows us as impatience with team members who are “slow to adopt.” It transforms into a tendency to form a working group rather than make a judgement call. It ends up in a search for someone to hold accountable when the transformation stalls.

The researchers call this “defensive organizing” and document its progression through four stages:

1. a comforting vision that doesn’t survive scrutiny,
2. a convenient foe (usually “resistant” employees or “slow” middle managers),
3. empty rituals of progress, and
4. a final reckoning that typically looks like restructuring or blame.

The Buffering Role That Breaks When You Are Not Okay

Here is what makes this more than a personal wellbeing issue.

Research on leaders and technological change is consistent: leaders serve as critical buffers between organizational anxiety and the teams they lead (Cao et al., 2021; Suseno et al., 2021). When employees are uncertain about what AI means for them, they look to their leaders, not primarily for information (that too), but for emotional support. How should I feel about this? Is this actually okay?

Leaders who have processed their own anxiety can hold that function. They can sit with uncertainty without transmitting it as panic. They can say “I don’t know exactly how this unfolds” and have it land as honesty rather than alarm.

Leaders who haven’t processed their own anxiety struggle to do this. And their teams can feel it. Not in what is said, but in what isn’t, in the rehearsed quality of the reassurance or in the forced enthusiasm.

And then many leaders are making it worse by removing themselves from the difficult human moments entirely. Dan Goleman, writing for Korn Ferry this week and drawing on a study of over 12,000 real AI interactions, found that personal and professional emotional support has become AI’s fastest-growing use case with therapy and companionship interactions doubling year-on-year (Is AI Eroding Your Emotional Intelligence?).

Leaders who use AI to draft difficult messages, prepare for sensitive feedback conversations, and mediate team tensions are bypassing the exact developmental discomfort that builds emotional capability.

The very presence employees need most during an anxious transition, a leader who can stay human in uncertainty, is being progressively outsourced.

What Actually Works

Here is the counterintuitive finding that emerges consistently across the research this week – leaders who honestly acknowledge the uncertainty build more trust than by projecting confidence. Actual honesty about not knowing what comes next is probably the best move for us leaders. Leaders who showed genuine openness to “grey areas” during the pandemic were significantly more effective than those who projected false certainty (Your Team Is Anxious About AI. Here’s How to Talk to Them About It).

Research on AI agent adoption found that explicitly acknowledging the limitations of the AI models, rather than just presenting its capabilities with confidence, increased employee trust and adoption rates (To Adopt AI at Scale, Employees Need to Trust Agents). IMD’s recent guidance to business leaders on navigating AI risk included a direct and rare recommendation: “admit uncertainty to employees about how work will change” (AI Leaders Are Pulling the Safety Alarm).

This is not just good for the team and the transformation project but also good for the leaders themselves. This allows them to objectively learn what the limitations of the technology are so that they can complement them with their and the skills of their teams.

This is the opposite of what most change management playbooks teach.

The MIT Sloan research (Fitzsimons et al.) identifies three practices that distinguish leaders who navigate this moment well from those who produce defensive organisations:

Courage:

Leaders who name their worry out loud as a practice do better. “I don’t have all the answers here. This is what I’m working through”, isn’t weakness. It’s the act of sensemaking, one that gives your team permission to be honest too. It also signals that the conversation is real, not managed. People can tell the difference.

Curiosity:

Leaders who keep a truth-teller, someone who can stress-test your thinking rather than just affirm it, do better. This is the exact opposite of what a typical AI assistant does (until trained better) and the difference matters enormously right now. Leaders who lean into their curiosity of what is working, who lean into the diversity of their team to understand what they are seeing, do better.

Care:

Leaders who visibly support the people who are exhausted do better. They do this not through some stated initiatives or programs, but through their presence and emotional support. The transformation is not just an organizational event, it is impacting real people, in real time, on top of everything else they are already carrying. Leaders who know when their people need a hug, or be heard or needed help and provided them what they need, do better.

What would Leaders Worth Following do Differently:

The executive I spoke with, the one who admitted, in a candid moment, that he worried as much about what AI was doing to him as to his team, did something that is rarer than it should be. He said the thing out loud admitting to what he was feeling. That’s the first step in understanding how AI is affecting us as leaders and to process it.

Most don’t. The leadership identity is built around having all the answers, projecting direction, being the person others come seeking guidance. Admitting anxiety feels like it conflicts with the role.

Leaders worth following are not the ones who project the most confidence about AI. They are the ones who stay the most human in its presence, who acknowledge what they don’t know, who resist the urge to perform certainty they don’t feel it, who show up and have the difficult conversations rather than delegating them to a tool.

Your team is not looking for someone who has it all figured out. They are looking for someone worth following through the uncertainty, through the liminality that we are collectively experiencing. That is a character question, not a competency question.

The true work begins not in the all-hands meeting you host, but in the honest conversation you have not yet had with yourself or with your team.

If this resonated, I’d love to hear what’s on your mind — reply or comment below.