Member portal

Leading in the age of AI

When the best idea in the room isn’t yours

Five shifts in how we lead now, and the human work underneath each one.

Michel Koopman, CEO and founder of CxO Coaching and a longtime senior operator, asks a question in a recent Fast Company piece that I find myself putting to leaders increasingly: what happens to your leadership when the best idea in the room isn’t yours?

He admits that the first time an AI system out-thought him on a resource decision, his ego wouldn’t let it make the call, so he made it himself. I appreciate that honesty, because I see the same reflex in capable leaders every week, and AI is going to test it far more often than any colleague ever did.

Here is how I look at it. For decades we have promoted people for their analysis, their speed, and their expertise, and AI is now doing much of that work well. What it cannot do is lead people, which means the part of the job that was always hardest to teach is quickly becoming the whole job. That is the lens I bring to Koopman’s five shifts, and in my experience each one comes back to the same place: the leader’s own clarity, self-awareness, and willingness to own the outcome.

Koopman suggests sorting the decisions you make repeatedly into two groups, letting intelligent systems inform the structured, data-heavy ones while keeping people accountable for those that carry ethical, reputational, or human consequences. I agree, and I would add that most leaders cannot do that sorting until they are clear about their own role.

In my work the starting point is always the five Key Responsibility Areas a leader truly owns, because a leader who cannot name those five will struggle to say which decisions belong to them, which belong to the team, and which can be handed to a system. Before I conclude that a leader is struggling, I diagnose the seat, since a role built on a vague mandate will defeat whoever sits in it, with or without AI.

Koopman asks whether we cling to hierarchy out of habit, and he is right that the real questions now concern who sets the rules, who audits the system, and who decides when to override it. The common corporate reading of AI is that we will simply need fewer managers. I read that differently, because an organization that halves its management layer has doubled the consequence of every manager left standing.

We are asking a tired and thinning layer of leaders to carry this change, which calls for choosing and developing them with far more care, not less.

That is a line I would like every leader I coach to adopt, along with his practice of naming out loud who watches the outputs and who owns what happens once someone acts on them. What I would add is that speed puts pressure on a leader, and pressure reveals whatever that leader has not yet examined in themselves.

The research of organizational psychologist Tasha Eurich, author of Insight, found that about 95 percent of people believe they are self-aware while only 10 to 15 percent actually are. A leader carrying that blind spot will either defer too quickly to the machine or, as Koopman candidly admits he did, override it to protect an image. Self-awareness and self-regulation are what allow a leader to hold the decision with confidence rather than ego. Having said this, I respect Koopman’s vulnerability here.

Koopman observes that human nuance is usually the first thing a new system drops, largely because it is harder to measure than efficiency. This is where I ask leaders to exercise what I describe as enterprise judgment, weighing how a decision made in their own corner lands on customers, peers, and the organization’s reputation, and feeling ownership for that reputation even when protecting it costs them locally.

Deloitte’s 2026 Global Human Capital Trends report makes the business case plainly: organizations taking a technology-first approach to AI are 1.6 times more likely to fall short of their expected returns than those taking a human-centric one. The return on the technology depends on the quality of the human system it lands in, and that system is built and sustained by leaders.

Koopman’s rule for his own team is that they automate analysis, not accountability, and he points to feedback that shapes someone’s career as a conversation no system should carry. I would widen that circle to include every moment when people are afraid, and AI has become a very real source of that fear for many of the people we lead.

Guiding a team through it takes a leader who can sit with other people’s anxiety without rushing to fix it, flee it, or paper over it, and that capacity has very little to do with being the best analyst in the room. This is emotional intelligence doing its real work, which is less about being pleasant and far more about being effective when the stakes are human.

Koopman closes by saying that leadership once meant knowing more than everyone else in the room and now means knowing exactly what you will hand off and holding the line on what you won’t. I would put that in the language I use with every leader I coach.