It is Thursday morning, and one of your directors has just produced a market analysis in an hour. Not long ago, that work would have taken the best part of a week. You thank them in front of the leadership team. Across the table sits a colleague who spent two days building her report the old way, and she says nothing. She is working out what you have just told the room counts as work. Before the meeting ends, she will have adjusted her behaviour accordingly.
Nobody in that room will ever raise it, which is precisely why it matters. Every boardroom I walk into is asking how AI can help the business move faster. Leaders want to automate more and prove a return on the investment. The bigger question is what AI is quietly doing to the relationships between your people. Almost nobody is asking it, and that is where the real risk is building.
The Debt That Does Not Appear In Your Management Accounts
Most leaders understand financial debt, and anybody who has sat near an IT function understands technical debt. Deloitte’s 2026 Global Human Capital Trends report names a third kind, which it calls cultural debt. It is the cost an organisation builds up when it neglects the behaviours and unwritten rules developing inside it. Like financial debt, it compounds slowly and quietly until you face a far bigger problem than the one you started with.
AI is accelerating the borrowing. Deloitte found that 42% of workers say their organisation rarely evaluates the impact of AI on people. Meanwhile, the same organisations track adoption in forensic detail. And 80% of leaders, managers and workers worry that colleagues use AI to look more productive than they are. No software update will fix that figure, because it describes a collapse in trust between colleagues.
Culture Lives In What Happens On A Thursday Morning
Every organisation runs two cultures: the one on the wall and the one people actually experience. Culture lives in who gets rewarded and who gets blamed when something goes wrong. It now also lives in how people use AI. If an AI recommendation turns out to be wrong, who owns it? When questions like that go unanswered, people answer them privately, each in their own way. That is exactly how the debt builds up.
Four Places The Interest Is Already Building
I set out five trust risks in the episode, and four of them appear in almost every leadership team I work with.
- Blurry ownership. AI can draft and recommend, but it cannot admit in a meeting that it got the call wrong. It can support a decision without ever owning the consequences. If your people do not know where that boundary sits, neither will your culture.
- Performed productivity. Researchers at BetterUp Labs and Stanford’s Social Media Lab call polished output with no thinking underneath it workslop. If you reward volume because AI has made volume cheap, your people learn that looking busy is what you value.
- Fairness. One person finishes in two hours what takes a colleague eight, so what exactly are you rewarding? Deloitte found that 65% of respondents believe their culture needs to change significantly because of AI. And 34% say culture is already getting in the way of their AI goals. If your AI strategy is struggling, you may have a cultural problem wearing a technology hat.
- Human connection. This is the risk that worries me most. People now ask a chatbot for help instead of a colleague, and each exchange looks efficient. Yet those small, disappearing conversations are where people build trust and notice when someone is struggling.
What A Horse Herd Knows About Coherence
My horses make that last point better than any report could. A healthy herd is a living social system rather than a group of individuals completing tasks. Every horse is constantly sensing and adjusting to the others, reading signals we would never notice. Take that connection away and you might keep a set of efficient individual horses. What you lose is the coherence that keeps the herd safe, and your leadership team works in exactly the same way.
The fifth risk, the erosion of judgement, is the one I would most like you to hear in full. It includes the moment I caught myself nearly falling into it with my own content.
Technology Magnifies The System It Enters
Put AI into a high-trust culture and it will magnify that trust. Put it into a fearful, siloed culture and it will magnify that just as efficiently. Your team will take their lead from you on this. If you hide your own use of AI, they learn secrecy. If you accept its outputs without question, they learn dependency.
Where To Start This Week
Begin by naming the questions people are already asking privately. Is using AI cheating, and what does good work look like now? Silence does not stop a culture forming; it simply means the culture forms without you.
Next, take one decision your team made with AI’s help last month and ask who owns the outcome. If the honest answer is “the system”, you do not yet have an answer.
Finally, identify the conversations that must stay human and protect them deliberately. If you do not design connection into your organisation, efficiency will gradually design it out.
So here is the question I left listeners with. Imagine I walked into your organisation six months from now and watched your people using AI. What would that tell me about what you truly value?
Listen to Episode 130 of Impactful Teamwork for all five trust risks and the two further actions I recommend. You will also hear why a 48-page AI policy does far less than a handful of principles people actually use.
Then join us at our next executive forum, where we work through exactly this question with leaders of scaling businesses and pressure-test what human leadership needs to look like alongside the technology. Register your place here.
If you would rather start with your own team, take the Entrenched Leadership Dynamics Diagnostic and find out which patterns are already shaping decisions in your business.
Show Notes
00:46 AI Shaping Culture
02:20 Cultural Debt Explained
05:16 Unwritten Rules Shift
06:54 Trust Risk One Ownership
08:59 Trust Risk Two Work Slop
11:05 Trust Risk Three Fairness
13:07 Trust Risk Four Connection
15:35 Trust Risk Five Judgment
18:05 Design Human Machine Partnership
19:39 Five Leader Actions
