Every organization I work with is asking the same two questions right now, though rarely out loud. When did we stop expecting leaders to be good people? And does it even matter, if AI can make better decisions anyway?
There was a moment when we decided character and competence were separable. Aristotle would have found this incoherent — for him practical wisdom was inseparable from goodness. To be rational was to be good.
The 20th century managerial revolution changed that. Management became a technique, clean of character and context. Teach the method and you don’t need the person; you just need a role at scale. A role you perform, not a person you become.
The competency movement of the 80s gave us behavioral frameworks, and an unintended result. Once leadership was codified into observable behaviors, you could coach the appearance of leadership without developing the character behind it. Performance frameworks, competency models, personal brand culture — all rewarding what is legible: communication style, decision-making, executive presence. You don’t need to know the work. You don’t need to be good. You need to perform management in ways others can evaluate. That’s how we built a competence theater, where the performance of authority replaced its substance.
We didn’t set out to exclude goodness; we selected for what was legible, and goodness rarely is: courage often looks like resistance, integrity doesn’t scale, neither survives a 360 survey.
Then AI arrived. It isn’t competing with leaders on execution but on authority. It does what organizations have rewarded for decades: it processes information, makes predictions, and projects confidence — faster, more consistently, and without ego.
The good news is that this creates a genuine opportunity to redefine what we select for. What remains is what was never legible: judgment under pressure, and relational integrity — whether people experience you as performing authority or as someone they trust. Most leadership pipelines aren’t developing this, or building the language to evaluate it.
We’ve been training people out of productive uncertainty and into performed confidence long before they reach leadership. By the time they have real influence, they’re skilled at the very behavior that makes them hard to trust at scale. The cost falls on the people below them: the exhaustion of being managed by someone you don’t trust, the loneliness of organizations where everyone performs competence at each other.
The disengagement this produced was measured for decades, and blamed on culture, communication, anything that didn’t implicate the system itself. What we were actually measuring was a trust deficit. Trust has one source: the person who proved, at personal cost, that they’ll do the right thing. Organizations that build around that won’t just have better cultures. They’ll have the one advantage AI can’t replicate: people who actually want to follow.