Most organisations I work with now have an AI policy of some kind. But what they don’t yet have is a settled way of actually leading through it, day to day. Navigating this shift requires a strong AI policy organisational leadership approach that focuses on culture, communication, and clear guidance.
The advice that’s stuck with me most from a course I’ve been on recently is this: treat your AI agent like a very smart graduate with an appalling memory. It’ll do good work, but you have to be very good at giving clear instructions, and you have to check everything you get back – not just whether the links work and the facts are right, but what assumptions it’s made, what biases have crept in, and what it’s quietly left out. That’s a useful frame for the practical relationship. What’s less settled is what it means for people, culture, and leadership in a team or organisation.
The pros of AI integration
- More time: People are getting hours back that used to go on repetitive tasks, drafting reports, data analysis and research.
- Faster first drafts: It helps with getting a half-formed idea onto the page – AI is a decent thinking partner for that.
- Lower barrier to trying things: Someone who’d never have attempted a data analysis, a funding application, or a bit of code, now can.
The cons and cultural pitfalls
- Less thinking together: Decisions that used to involve a conversation now sometimes involve one person and a chatbot.
- Speed without space: Things move fast and there’s little room to ask what that speed is costing.
- Uneven confidence: Some people are using AI well and saying nothing about it, while others are avoiding it, worried about breaking a rule or fearful that proving their efficiency might cost them their job.
What I’m sensing underneath both lists, is that AI is reducing human interaction, in a working environment that’s already hybrid or virtual and where relationships aren’t being built and maintained the way they were in the workplace before 2020.
I believe work gets done better and faster when people know, like, and trust each other. If that’s right, I wonder whether AI is quietly disconnecting people from each other, and whether the lack of conversation and speed around decisions is also disconnecting people from the work itself.
Trust matters
Trust isn’t built through big gestures. It’s built in the small, slow moments: someone talking through a half-formed idea, a team disagreeing openly about a decision, a colleague admitting they don’t know something and someone else helping them work it out. Those moments need space, and space is exactly what’s being squeezed. They’re also the behaviours that make up psychological safety – the sense that you can speak up, ask questions, admit you don’t know something, and challenge an idea without fear of looking foolish or getting it wrong. When those moments disappear, so does the safety that goes with them.
AI is now part of the team
This is a subtly different territory for leaders, and it needs something different from what’s worked before. AI is now part of the team, in the sense that it shapes behaviour and decisions whether anyone’s named it or not. Leading that well isn’t a matter of writing a better policy or running a training session on the tools, (although AI literacy still matters and shouldn’t be skipped). It’s about helping your team slow down enough to think for themselves, keep building the trust that gets lost when the answer arrives too fast, and use AI well – rather than just often.
The skills this calls for are largely the ones I’ve always trained: asking better questions, checking assumptions, listening, making space for people to think out loud, and noticing when a decision has been made too fast to be a good one. AI hasn’t replaced that skill set. It’s just made it more necessary than it’s been in years.
If working with AI in your team or organisation is something you’re adapting to, I’d like to hear more about what you’re seeing. I’m setting aside time for short conversations with leaders navigating exactly this. You can grab a slot here.