Leading Through the AI Fog

What leadership actually requires right now, when the long-term plans are no longer available.

You are not being asked to figure out AI.

You are being asked to lead your people through a stretch of uncertainty most leaders have never operated in before. And you are being asked to do it without the long-term plans you used to be able to make.

Those are different jobs. The first one cannot be done. The second one can. And it asks something of you that the first never did.

Most of the conversation right now is happening at the wrong altitude. The pressure on founders and senior leaders is to develop better AI strategy. Sharper roadmaps. Clearer vision. The implicit promise is that with enough work, the leader can produce certainty about where the company is going and what AI means for it.

That promise cannot be kept. Not because leaders are not working hard enough. Because the conditions for that kind of certainty no longer exist.

The conditions have changed

Toby Stuart at Berkeley calls this the AI fog. His argument is that AI's rapid advance has created limits on leaders' visibility into the future, and that these limits are challenging the criteria leaders have always used to make long-duration commitments. “

Staring into this fog,” Stuart writes, “leaders will be tempted to trade the potential future gains from skyscrapers and railways for the temporary utility of tents and bicycles.”

His point is that the gap between risk and uncertainty is the heart of what has changed. Risk is when you can assign probabilities and price a bet. Uncertainty is when the probability distribution itself is unknown. We are now operating in the second condition, not the first, and most of the mental machinery of leadership was built for conditions we are no longer in.

The leader who tries to make decisions today the way they were making them before covid is not failing. They are using instruments that were calibrated for a future that no longer behaves the way it used to. Covid broke a set of stability assumptions that had held for decades. AI is the second wave on top of that. We have been operating in elevated uncertainty for six years now, and the second wave is making the first impossible to ignore.

What this asks of the leader

The honest response to fog is not to pretend it is not there. It is to lead inside it differently.

Practically, this means shorter horizons. Smaller commitments. Decisions that buy you information and the right to follow on rather than locking you into bets that are too large to revise. Stuart calls this optimising for the unknown. In a scaling company it looks like running quarterly rather than five-year plans, making the next decision rather than the next ten, and being honest with your team about what you can and cannot see.

This is the operational shift. It is the easier half of the work.

The harder half is what happens inside the leader while they do this. Operating in fog is psychologically demanding. The leadership instincts most founders developed are built around producing certainty for their teams. Knowing where the company is going. Pointing the way. Holding the vision. None of that is fully available right now in the same form. And the leader has to learn to lead anyway, without the props they used to lean on.

That is the part that is genuinely difficult. And it is where most of the discourse stops short.

What people need to function

Humans are not designed to operate in pure uncertainty. We are also not designed to operate in pure stability. We need both, in balance. Stability gives us ground to stand on, routine, the predictability our nervous systems rely on to stay regulated. Uncertainty gives us the novelty, challenge, and growth that keep us engaged and developing. Push the balance too far in either direction and the system breaks down. Too much stability and we rust. Too much uncertainty and our nervous systems shift into chronic stress, narrowing our focus, eroding our judgement, and quietly draining the capacity we need to do good work.

Most teams in scaling companies are currently being asked to operate inside more uncertainty than the system can comfortably absorb. AI is changing what their jobs might look like in two years. The middle management layer is being squeezed. The junior pathways are being collapsed. The technology they are being asked to adopt is changing month to month. None of this is going to settle soon.

The leader's job in this moment is to provide the stability that lets the team function inside the uncertainty. Not to remove the uncertainty. Not to produce false certainty about a future nobody can see. To be the source of steadiness from which the team can do the work of facing what is actually happening.

This is the actual reframe. The leader is no longer the person with the longest line of sight. They are the person providing the secure base from which the team can keep operating well while the ground keeps shifting.

When the balance breaks

The cost of leaders failing to provide this stability is now being measured. Recent research by Guy Champniss surveyed more than 1,200 employees across the US and UK and named six forms of psychological strain that accumulate when AI is rolled out without the human conditions designed around it. Cognitive load, autonomy loss, competency erosion, weakened connection between people, credibility anxiety, identity threat. Junior employees carry the highest scores, with rates significantly above their more experienced colleagues.

These are not symptoms of bad employees. They are symptoms of teams being asked to operate inside too much uncertainty without the stability that makes uncertainty workable. The cost shows up in the people, but it is generated by the conditions.

The leader who recognises this stops trying to fix it at the employee layer. The fix is not a training program or a wellness initiative. It is the leader providing the steady presence that lets the team's nervous systems stay regulated enough to do hard, uncertain work.

What that actually looks like, week to week

The leader who provides this kind of stability is not stoic. They are not performing calm they do not feel. They are doing three things consistently, in small ways, week after week.

The first is examining what their own role actually requires. What does the work of leading this company look like this week, given that some of what I did last quarter is now starting to change. Where is AI making part of my old job redundant. Where is it creating a new responsibility I have not yet named. The leader who keeps asking this question stays current with what their role actually is. The leader who does not keeps running the role they had two years ago, and the gap between what the company needs and what they are providing keeps widening.

The second is asking what the team needs from them this week. Not strategically, in a quarterly review. Practically, in the moment. What is the team carrying that I have not noticed. Where are they running ahead of me and needing me to catch up. Where are they waiting for me and I have not yet shown up. The leader who keeps asking this question meets the team where they actually are. The leader who does not keeps offering what they think the team should need.

The third is asking how they are helping their people stay steady through ongoing uncertainty. Not with false reassurance. With honest presence. Acknowledging openly what they do not know. Anchoring the team to what is genuinely stable underneath the changes, the work that matters, the relationships that matter, the way they treat each other. Normalising the discomfort of not having a clear picture, so the discomfort becomes shared rather than private. Making space for the team to bring what they are noticing about AI in their own work, rather than performing as if the leader already has the full picture.

These three questions are small. They do not solve AI strategy. They are the discipline of holding the balance, in the conditions that actually exist, week to week.

What this requires inside the leader

The leader who can do this is the leader who has done some of their own regulation work. Who has examined their own anxiety about AI. Who has stopped performing certainty they do not feel. Who has accepted, at some level, that they cannot see further than they can see and is at peace enough with that to lead from the peace.

This is the part most leadership writing avoids. The stability the leader offers the team comes from the stability they have built inside themselves. What the leader cannot hold inside themselves they cannot hold for their team. The team can read the gap between performance and presence at twenty paces.

This is not a call to therapy. It is a recognition that the work of leading through opacity is inner work as much as it is outer work. The leader who has not faced their own discomfort about not knowing the future will keep generating that discomfort for their team, no matter how good their operational shorter-horizon plans are.

Where this stops being one leader's job

There is a limit to what a single leader can hold for an organisation. Even the most regulated, attentive, present founder cannot personally be the source of stability for a hundred people, every week, in every part of the company. The model where stability flows from one person at the top is the same operating model that produces the bottleneck the previous issues have named. It cannot scale.

The work, eventually, is to develop the team's capacity to hold each other steady. To build the conditions in which people across the organisation provide the stability for each other that the leader cannot provide alone. That is a different kind of work. It is the layer where strategy meets the team and where the conditions for sustained performance are actually built.

How that work gets done is the next issue.

What stability are you providing for your team this week that is coming from your own regulated presence, and what are you providing that is performance you cannot sustain?