The Mona Lisa Smile

If you haven't seen it already, Ethan Mollick's recent piece, Management as AI superpower is well worth a read.

His core argument is simple: if "delegation is the new prompting," then the skills that matter most when working with AI agents are good management fundamentals — knowing what good looks like, setting clear goals, evaluating outputs, and giving useful feedback. The people who will thrive, he argues, are those who can direct AI clearly enough that it delivers. The much-maligned middle manager may have hope yet.

This resonated strongly. In the last few months, AI has hockey-sticked from "this might be real, we need to take this seriously" → "how do we not get stuck in pilots?" → "the capabilities have already moved again — and look what it just made." We've talked about technology as a boardroom issue for years — this time is different. In many organisations, it feels existential.

And yet, as I took Mollick's argument into discussions with different leadership teams, it became clear that management skills are, at best, only half the story.

Two things became clear. 1) Just as AI demands stronger management skills, it also demands stronger human leadership capability. 2) AI creates a sharper distinction between when each is needed — demanding what I've started calling bi-lingual leadership.

The Mona Lisa Smile

Here's something I've seen a few times recently. A leadership team is discussing what AI might do for their business. People begin to see the possibilities — the rapidly expanding capabilities, the potential to radically scale and accelerate. The CEO is getting excited: imagine, we could have an agent that does X automatically. Someone jumps in, builds on the idea. The suggestions are really good — because X is exactly what that person, or their team, spend 80% of their time doing today. They finish, smile, and the conversation continues.

I look at that person and wonder, what do they really think? How are they really feeling?

It's a bit of a Mona Lisa smile.

And that smile contains a leadership problem, not a management one.

The Other Half

Management is about execution — planning, coordinating, optimising. Leadership is about judgement — priorities, decisions, alignment. AI raises the bar on both. But it also makes the distinction between them harder to ignore.

Leadership has been defined in many ways, but two elements keep coming up at the moment. The first is direction: the ability to set a course, to articulate a destination that others can orient around. The second is people: the ability to build trust and belief in that direction.

Direction-setting — purpose, navigating shortened lifecycles, VUCA, the total societal impact that AI supercharges in any organisation — is a discussion for another day.

The more urgent conversation is about people. Leading people. That ambiguous smile.

Every week brings new predictions about job displacement. For a significant number of people today, what they are witnessing is not just change. It is existential concern. About relevance. About contribution. About what their life looks like in the near future. And no-one is feeling like they have loads of free time to write music and wait for their robot-enabled universal basic income. In fact, AI — like all technological advances before it — has not reduced work demand, it has intensified it.

Covid felt big, but in retrospect the challenge was to change how we did everything — the how. AI is challenging the what, the who, and in some cases the why. That is a different order of disruption.

AI makes it easier to produce outputs, run processes, and execute at speed. But it does not help address the single most important determinant of success and failure — how people feel.

This is often called 'the soft side' — spoken in a tone that implies it is of lesser importance than the real work of getting hard things done.

We know the opposite to be true. Google's Project Aristotle (measuring 180 teams from 2012-2016) found that psychological safety — feeling safe to take risks and be vulnerable — is the most important factor in high-performing teams, outweighing individual expertise. It is well established that people are the difference between successful and unsuccessful change. That is even more true when the change feels this large, this fast, and this personal.

In parallel with reading Mollick's article, I recently read an advance copy of a forthcoming book Leadership in Tune, by Ciarán Casey. Casey's central idea is that leadership is not about authority or control — it's about the relationship between people. It emerges amongst willing parties, through understanding, listening, and building a bridge between where people are and a shared direction.

In ordinary times, these insights are important. In the current moment, they feel critical.

Relational leadership — the capacity to meet people where they are, to hold the human alongside the strategic — is not a soft skill complement to the real work. In the AI era, it is the real work.

The Third Half — Bi-Lingual Leadership

Yes, three halves. But "that's only one third of the story" doesn't quite work either.

We all know management does not equal leadership. But how often is this really true outside the classroom, the books and the OD slides?

In most organisations, we have been getting away with a rough mush of the two.

The implicit model is a Venn diagram with a big overlap. The difference matters to academics, OD consultants and People & Culture, but functionally it has been comfortable enough to ignore. Most people get their first leadership/management role off the back of technical competence. Friday you're one of the team; Monday you're the boss. From there, an ad hoc collection of 'leadership' and 'management' training.

AI is pulling those circles apart. What constitutes effective management — precise task definition, structured delegation, evaluation and iteration at scale — is becoming more exacting. What constitutes leadership — judgement, direction-setting, trust-building — is becoming more exposed and more acutely needed. The overlap is shrinking.

What is needed is the ability to operate in two distinct modes, and to move between them with intent — like being fluent in two different languages, and knowing which one the conversation requires.

One mode is precise and analytical: defining work clearly, deploying it across people and AI, evaluating outputs, and iterating quickly. The other is relational: building trust, setting direction, staying with people through uncertainty. Establishing conditions where people feel valued and engaged in something they are proud to be part of — something they want to tell their friends and family about.

Leaders need to develop these not as a blended default, but as a deliberate switch — a conscious shift from one mode to the other as the situation demands.

For shorthand, I've started to call this bi-lingual leadership.

The implication is significant. We need an uplift in management capability — directing AI clearly, evaluating outputs, coordinating across systems that move faster than any team previously could. We need an equal uplift in leadership capability — purpose-setting, relational depth, the capacity to hold people through genuine uncertainty. And most importantly, we need to develop the judgement and agility to move fluidly between the two.

The organisations that succeed will not be those that convert the most people to AI agents. They will be the ones that can bring their people on the journey — sharpening the precision and the humanity, and developing the judgement to know which the moment requires. Where goals are shared, where people are engaged rather than afraid, AI compounds the gains. Those are the organisations that will accelerate, attract the talent they need, and take a leadership position in their markets.