The fastest way to miss the point

In a recent LinkedIn post, I shared an observation that I’ve been pondering on.

A few weeks ago, in one of our leadership simulations, something happened that stayed with me. The exercise wasn't complicated but the shortness of time and overwhelm of information made it feel more complex than it actually was. A team facing an ambiguous, human problem, the kind with no clean answer, only judgement calls and trade-offs. It's designed to surface how people behave under uncertainty; what they notice, what they ignore, how they bring others along. Simulations, in years of us running them, have never been about producing a correct output.

One participant opened a chat window and asked AI to solve it.

Not maliciously. Not lazily, even. Quickly, instinctively, the way you'd reach for a calculator. Within a minute they had a tidy, plausible answer. What they didn't have was any insight into their own behaviour, which was rather the entire point of the exercise. They solved the simulation and missed the moment it was built to create.

I don't think that participant did anything wrong. I think they did something completely understandable, and that's what makes it worth sitting with.

Everyone's talking about AI. Fewer people are talking about leadership.

There is no shortage of content right now on what AI can do, what it might replace, how fast it's moving. Which means our next natural question isn't really about the technology at all. It's what is required of leadership, to support people and organisations that are using it. Not the tool literacy. The judgement. The discernment about when reaching for AI is exactly right, and when it quietly removes the human thinking a moment actually needed.

That gap is where our attention has been this year. We asked a group of postgraduate students at the London School of Economics - a partnership we value and look forward to working together each year and always come away from with more than we expected - to look specifically at how leaders can provide better support in an AI enabled future. The research was carried out by Natalie Delgado and Josephine Pfeilstetter, and the full findings are coming in the weeks ahead. There's a lot in there worth sharing properly. But one thing surfaced early enough, and clearly enough, that I wanted to write about it now rather than wait.

Capability on its own isn't the story

The instinctive framing is that this is a leadership development problem. Teach leaders the right skills, the right fluency, the right judgement, and the rest follows. The research pushes back on that, gently but firmly. Capability matters, but on its own it isn't enough. What actually determines whether leaders behave well around AI is whether that capability is met by the right conditions inside the organisation. Without them, you get leaders trying to do the right thing inside a system that was never built to support it.

Behaviour is environmental, leaders included

This isn't a new idea for us, it's one we've built a good deal of our thinking on over the years, and it has real weight behind it beyond our own observation. Kurt Lewin, one of the founding figures of social psychology, put it simply back in the 1930s: behaviour is a function of the person and their environment, not the person alone. Decades of research since has only reinforced it, including the well documented tendency we all have to explain someone's behaviour by their character while overlooking the situation that produced it, noticing the person and missing the conditions around them. It's also the logic underneath the COM-B model, developed by Michie, van Stralen and West at UCL, which treats opportunity, the physical and social conditions someone is operating in, as being just as load bearing on behaviour as capability or motivation. Take any one of the three away and the behaviour tends to collapse with it.

Give people the right conditions and they tend to rise to them. Give them the wrong ones and even strong capability struggles to show up consistently. It's why so much of our work with clients has never stopped at developing leaders in isolation. It's always been about the conditions those leaders then create for everyone around them to perform against the outcomes the business actually needs.

What the LSE research argues is that naturally, this same logic applies to leaders too. Leaders are humans first. If behaviour is environmental for the people leaders are responsible for, it's environmental for the leaders themselves. Which means organisational conditions can't sit downstream of leadership development, arriving once the capability building is done, as a nice to have. At the very least they need to be built alongside it. Ideally, they're in place to enable leadership behaviour develpoment. Ask a leader to hold better judgement, better fluency, better relational intelligence, without the conditions to support it, and you're asking them to be the exception to a rule you'd never apply to anyone else in the business.

In the LSE research, four of these pairings stood out.

Fluency needs infrastructure. Leaders should build real fluency in AI, understanding what the tools actually mean and do, and to keep adapting as the landscape shifts under them weekly. Reasonable. But fluency doesn't develop in a vacuum. If the organisation hasn't built the resources and infrastructure to learn from, that expectation just becomes pressure without a path.

Judgement needs governance. Leaders create value when they exercise real discernment, not just about whether AI's output is good, but about whether using it is appropriate in the first place. Without governance sitting alongside that, discernment collapses into personal opinion. Everyone ends up drawing their own line, and no two lines match.

Vision needs alignment. Leaders help people make good decisions when they orchestrate AI use with a clear strategic intent behind it. Without genuine strategic alignment from the organisation, that vision is something each leader has to invent for themselves, usually badly, usually inconsistently.

Relational intelligence needs culture. This is the one that sits closest to how we think at LIW. Leaders are expected to protect trust, preserve real human connection, and watch carefully to counter the moment thinking gets quietly outsourced rather than supported. They're expected to hold enough psychological safety that people can experiment and get things wrong. None of that is possible if the surrounding culture doesn't hold it too. A leader can try to build safety in a room. They can't build it against the current of the organisation around them.

Back to the room

Which brings me back to that participant.

They weren't short on capability. They were quick, resourceful, comfortable with the tool. What was missing wasn't in them at all. It was the surrounding condition that would have made pausing, rather than solving, feel like the obvious move. Nothing in that moment signalled that the behaviour was the point. So they did what most of us would do. They reached for the fastest route to an answer.

That's the part of this research I keep returning to. There’s a danger that we can assume the leadership development conversation is a capability conversation. Teach the skill, run the workshop, tick the box. What LSE's work is pointing to is something less comfortable and more interesting: that leadership capability and organisational conditions aren't two separate workstreams. They're interconnected. Treating them as separate is probably why so much of this work fails to stick.

We'll share the full research from Natalie and Josie at LSE in the coming weeks, and I think it's genuinely some of the most useful thinking we've had back from this partnership. For now, though, I'd rather leave you with the question it left me with.

Next time someone on your team reaches straight for the tool, what does that actually tell you? About them, or about the leadership capability you've built around them?