Can Your Culture Carry AI?

Series: What Can Your Culture Carry? · Part 1 of 4

Someone on your team used AI to write something a client read last week.

A proposal. A scope note. A safety summary. An email that set an expectation you will now have to live with.

You did not see it before it went out. You did not check the judgment inside it. And that is not a policy failure. It is a preview of a question most growing companies have not asked out loud yet: what is our culture actually able to carry, now that everyone has a machine that sounds confident?

I work with technical businesses in engineering, construction, manufacturing, and energy. Capable people, real pressure, thin margins. And here is the pattern I keep noticing. AI did not arrive through a strategy deck. It arrived quietly, one person at a time, in the gap between “we should probably talk about this” and “everyone is already doing it.”

That gap is where culture lives.

AI is a culture problem wearing a technology costume

Most leadership teams are treating AI as a tools decision. Which licence, which platform, which policy. Those matter. But they are not where the value is won or lost.

The evidence is getting hard to ignore. In PwC’s 29th Global CEO Survey, released at Davos this January, more than half of chief executives said they had seen neither higher revenue nor lower costs from their AI investments in the past year. Only about one in eight had captured both. The tools are extraordinary. The returns, for most, are not there yet.

Why? Not the model. The operating culture underneath it.

Deloitte’s 2026 Global Human Capital Trends report puts numbers to what I see in the field. Sixty-five percent of organizations believe their culture needs to change significantly because of AI. Forty-two percent of workers say their organization rarely evaluates the effect of AI on people. And ninety-three percent of AI investment is going into technology, while seven percent goes into the humans expected to use it well.

Read that last one again. We are funding the machine and starving the judgment.

This is the Capacity Gap in its newest form: the distance between how fast the work is changing and how ready the people are to lead through it. AI did not close that gap. It widened it, and it made the gap visible faster than any training calendar can keep up with.

What breaks under AI

When a culture cannot carry AI, it does not fail loudly. It frays quietly, in five places.

1. Decision rights. Who decides what responsible use looks like? If the answer is “nobody has said,” then everyone is deciding privately, and you are running twelve different AI policies at once without knowing it.
2. Judgment. AI is fluent, and fluency reads as competence. When a tool sounds sure, a tired person stops checking. The question a strong culture asks out loud is simple: how do we know when this is wrong?
3. Trust. Deloitte found that eighty percent of leaders and workers worry their colleagues are using AI to look more productive than they are. That is not a technology statistic. That is a trust statistic. When people use AI in the quiet, they are telling you something about whether it is safe to use it in the open.
4. Accountability. When AI drafts it and a person sends it, who owns the outcome? If that is unclear before something goes wrong, it will be brutal to sort out afterward.
5. Expectations. Here is the one nobody names. AI rarely removes work. It quietly raises the bar for what “done” looks like, until your best people are carrying a standard no one ever agreed to.

None of these are tool problems. Every one of them is a culture problem. And culture, in a growing business, is simply the accumulation of what leaders decide, notice, reward, and tolerate.

Three moves for this week

You do not need an AI strategy offsite to start carrying this well. You need three conversations.

1. Name what responsible use looks like, in your context, on one page. Not a legal policy. A working agreement: where AI helps here, where it does not touch the work, and what always gets a human check before it leaves the building. Decision rights first. Everything else follows.
2. Put judgment back in the loop. Pick the two or three places AI now touches client-facing or safety-critical work, and name who verifies before it ships. You are not slowing the tool down. You are keeping a human accountable for the outcome.
3. Move it from quiet to open. Ask your team, without a hint of a trap in your voice, where they are already using AI and where they are unsure. The goal is not to catch anyone. It is to make the real practice discussable, because you cannot lead what people are hiding.

When the two largest research houses in the world diagnosed the AI returns problem this year, they did not land on better models. PwC pointed to foundations, governance, and culture. Deloitte pointed to clear decision rights and trust. That is the whole argument of this series, arriving in someone else’s data: the constraint is rarely the technology. It is what the culture can carry.

How to support your people:

This is not a reason to fear AI, and it is certainly not a reason to slow your people down. The capability inside your organization is real. Your people are more ready than you think, if you give them clarity instead of a vacuum.

But clarity is a leadership act, not a software setting. A culture that can carry AI is one where decisions sit at the right level, judgment is checked without shame, and people can be honest about what they are actually doing. Build that, and AI compounds your strengths. Skip it, and AI compounds whatever was already thin.

Growth does not create a culture problem. It reveals what your culture can carry. AI just turned up the speed.

One next step. If you want to see where your culture is strong enough to carry this and where it is already straining, the Culture Capacity Assessment takes about ten minutes and reads five capacities that decide what your organization can hold: decisions, accountability, leadership, adaptability, and resilience. It is a diagnostic, not a pitch.