If you lead a business right now, AI isn't a future decision you get to make deliberately. It's already inside your organisation — in the tools your team chose for themselves, in the shortcuts they've quietly built into their week, in the way your managers are drafting emails and briefs and performance reviews. You didn't necessarily roll it out. That's exactly the risk.
That gap — between adoption and competence — is where the real business risk sits. And it isn't only a leadership problem. Your managers, your senior staff, the people running your client relationships: they're using AI on their own initiative too, without a shared standard for how, without anyone checking whether it's helping or quietly costing you something. Our latest Performance Blueprint webinar took on that question directly. Matt Elliott and Susan Pearse sat down with Mary Gourley — over a dozen years in an executive role at Google — to unpack what's genuinely at stake for a business, not just what's technically possible with the tools.
The upside is real — this isn't a case against AI
Used well, AI is already solving problems that were previously too slow or too costly to catch: Australian banks and telcos running real-time fraud prevention that flags a suspicious call before money moves out of an elderly customer's account; retailers catching mispriced items at the till through image recognition; CSIRO predicting fire behaviour so evacuation teams spend their time keeping people safe instead of manually mapping a fire's likely path. Handled with intention, AI buys your business real time and genuinely reduces risk. That's the case for engaging with it seriously — not the case for leaving it unmanaged.
But left unchecked, here's what it costs you
The cost doesn't arrive as a single, visible event you can point to and fix. It arrives gradually, and it's easy to miss while it's happening. Mary named the pattern plainly: the biggest risk of AI isn't machines replacing people. It's people starting to act like machines. Spend enough hours a day issuing short, transactional prompts to a chatbot, and that register starts leaking into how your people message each other. Crisp becomes brusque. Efficient becomes cold. And because a leader's communication sets the standard for the whole business, that drift doesn't stay personal — it becomes cultural.
You start to emulate that. You become very crisp and brusque, and you lose the human connection — because that's how you're putting commands into an interface.
Play that forward twelve to eighteen months. A more disconnected workforce doesn't announce itself — it shows up as engagement scores that quietly slide, as the culture work you've invested years in starting to erode without an obvious cause, as your best people becoming harder to retain for reasons nobody can quite name. By the time it's visible in the numbers, you're no longer managing an AI adoption question. You're managing a culture and retention problem, and the root cause is buried under a dozen other explanations. It is far harder — and far more expensive — to rebuild trust and connection than it is to protect it while you still can.
Three signs it's already starting
Mary named three specific, observable patterns worth watching for inside your own business — not abstractions, but things a leader can actually see in how their team works day to day.
- The AI-to-AI loop: One person uses AI to draft a memo. The recipient uses AI to summarise it back. Neither has actually read or written the thing — but it gets called collaboration. It isn't. Writing something yourself commits it to memory; outsourcing both ends of the exchange means nobody has really connected at all.
- The death of the water-cooler conversation: The half-formed question someone used to lean over a desk and ask a colleague now goes to an LLM instead. The information comes back faster — but the mentorship, the war stories, and the relationship-building that used to happen inside that exchange simply vanish.
- Treating teammates like prompt boxes: Hours spent typing commands into a chatbot don't switch off when someone opens a chat window to a colleague. The same blunt, stripped-down register carries over — and repeated across a business, it drains the empathy out of everyday communication.
Automate the task. Not the responsibility.
Matt's closing summary of the session is the sentence worth carrying into your next AI decision as a leader: automate the task, not the responsibility. AI can take a task off your plate. It cannot take the responsibility to connect with your people, read the room, or make the judgement call that only a human in that specific moment can make. That's the discipline that captures the upside without absorbing the cost: reinvest the time AI gives back into your people, your thinking and your own capacity, instead of letting it get swallowed by the next task; treat prompting as a skill worth building properly across your team, not a one-line afterthought; and protect human connection deliberately, on purpose, because it's the one thing AI was never built to replace — and the one thing that's hardest to win back once it's gone.
This is already happening in your business, whether or not you've decided how you want it to. If you'd like to talk through what that looks like for your team specifically — how to capture what AI genuinely offers without paying for it in culture and connection — we'd welcome the conversation.