AI Isn’t Changing Your Culture. It’s Revealing It.

What a debate about AI taught me about the gap between what we say and what we do.

I argued something in a debate recently that I haven’t stopped thinking about since. The debate itself almost doesn’t matter. What matters is the realization it forced me to articulate, which is that most companies think AI is going to change their culture. I don’t think that’s true. I think AI is going to reveal exactly what their culture already was.

But before I get into why, I want to back up and ask a question that almost never gets asked directly in these conversations. What actually is culture?

Because if you can’t answer that clearly, the rest of this doesn’t mean anything.

What Culture Actually Is

Culture isn’t the values on your website. It isn’t the mission statement framed in the lobby or the line in the handbook about “moving fast and being kind.” Those are aspirations, and aspirations are not the same thing as culture.

Culture is the answer to one question: what actually happens around here?

What questions get asked, and who’s allowed to ask them. What happens when someone challenges a decision. Who gets the benefit of the doubt and who doesn’t. What gets rewarded when nobody’s watching, and what gets tolerated even though it shouldn’t be. That’s culture. It’s not what you say. It’s what you do, repeatedly, until everyone in the building has learned the real rules without anyone writing them down.

This matters enormously for the AI conversation, because most organizations are about to find out that their stated values and their operating culture were never quite the same thing. AI didn’t create that gap. It just makes it impossible to miss.

The Argument I Made

The debate I was part of centered on a single question: should companies adapt their culture to fit how AI works, or should AI be expected to fit into the culture that’s already there?

I argued the company should adapt. Not because AI deserves special treatment, but because work itself has changed in ways that most operating cultures were never built to handle.

Here’s what I mean. AI moves fast, crosses every functional boundary without asking permission, and is confidently wrong on a regular basis. It doesn’t flag its own uncertainty the way a junior employee might, it just produces an answer with total conviction, right or wrong.

Most company cultures were built for the opposite conditions. Stay in your lane, defer to whoever has the most seniority or the most tenure, move at the speed of consensus. Trust the output of the person, or now the system, that has historically been right before.

None of that holds up well against a tool that’s fast, cross-functional, and wrong in ways that look exactly like being right.

What This Actually Exposes

Here’s where the debate got interesting, and where I think the real argument lives. AI isn’t introducing new problems into organizations; it’s taking problems that were already there, problems that the existing culture was accommodating, and putting them under a spotlight nobody can look away from.

Poor judgment doesn’t show up only when someone makes a bad call anymore. It shows up when someone accepts an AI’s bad call without questioning it, because questioning things has never been safe in that building.

Weak communication doesn’t just slow down a project. It means nobody’s translating what the AI produced into something the team can actually act on, because translating has never been anyone’s defined job.

Low trust doesn’t just create friction in a meeting. It means a junior employee who can see the AI’s output is wrong stays quiet, because speaking up has never gone well for anyone at their level before.

Unclear accountability doesn’t just create confusion about deadlines. It means nobody actually owns whether the AI’s recommendation gets checked before it goes out the door, because ownership was always a little fuzzy to begin with.

None of this is new. Your culture had these patterns long before anyone bought an AI license. What’s new is that there’s no longer anywhere for the pattern to hide. AI moves too fast and touches too much for an organization’s old habits of absorbing dysfunction to keep working.

The Question Underneath the Question

There’s a specific tension I raised in the debate that I think gets to the heart of this whole argument.

For decades, most organizations have relied on a simple assumption: the person with the most experience is usually the person with the most knowledge; that’s how authority has worked. You earn experience, experience earns credibility, and credibility earns influence.

AI has complicated that equation almost overnight. Expertise and authority no longer automatically sit in the same person.

For the first time in most of our careers, the most experienced person in the room might need the least experienced person to show them how something works. The senior leader brings years of judgment, built from seeing what good and bad decisions actually cost. The newest hire may bring far greater fluency with the technology. Neither capability is enough on its own.

The real question isn’t whether senior leaders should learn AI - of course they should.

The question is whether your culture allows expertise to move in both directions. Can a junior employee teach a senior leader without it feeling uncomfortable? Can a senior leader admit they don’t know something without worrying it undermines their authority? Can both people leave that conversation with more credibility than they started with?

Because the organizations that adapt fastest won’t necessarily be the ones with the smartest AI. They’ll be the ones where expertise can move freely, regardless of hierarchy.

That leads to the question I think matters most. Does your culture make it normal, not just technically permitted but genuinely normal, for someone to say “I think the AI got that wrong”? Or does it reward agreement, because disagreement has always been treated as a risk rather than a contribution?

If the answer is the second one, you don’t have an AI problem. You have a culture that was already optimized for compliance over judgment, and AI just gave you a faster way to find that out.

Why This Isn’t an AI Adoption Problem

I want to be direct about where I land on this, because I think a lot of organizations are solving the wrong problem entirely. They’re treating this moment as a technology rollout. New tool, new training, new rollout plan, done.

I think it’s a leadership capability problem wearing a technology costume.

The technology scales instantly. You can roll out an AI tool to five thousand employees in an afternoon. Leadership capability doesn’t. It’s built slowly, through people being coached, challenged, and given real feedback over time, by leaders who were themselves developed the same way. Most organizations have spent the last decade underinvesting in exactly that kind of development, and now they’re discovering the bill is due at the worst possible moment, right when the demands on every leader have multiplied.

That’s why so many companies feel like they’re moving faster without actually moving better. The output is increasing, but the judgment guiding that output hasn’t kept pace.

What I Actually Believe

AI won’t replace judgment. But it will expose, immediately and at scale, every place judgment was already missing.

The organizations that come out ahead of this moment won’t be the ones with access to the best models. Model access is not a moat anymore; everyone will have roughly the same tools within a couple of years. The organizations that come out ahead will be the ones whose leaders know how to use those tools well, which means knowing when to trust the output, when to challenge it, and how to build a team where challenging it is not just allowed but expected.

That’s not a technology investment. That’s a culture investment, and it’s one of the biggest investments most organizations still aren’t making, at exactly the moment they need it most.

AI raises the ceiling on what’s possible. Whether your organization actually reaches that ceiling has very little to do with the AI, and everything to do with the leadership underneath it.

AI was never going to change your culture. It was always going to reveal it.