People with a low tolerance for uncertainty describe AI as "a threat" three times more often than people with a high one, and see it as an opportunity at a third of the rate: 24% against 69%. Same technology, opposite experience. It's not easy to experiment when you're anxious.

It's not easy to experiment when you're anxious.

If you've come across Uncertainty Experts before, you've heard us talk about the thought versus felt gap. It was our original contribution to the uncertainty debate, and we're humble-braggingly proud to see how far it's been adopted. In 2026, the most uncertain year ever (again), the gap between how we think (usually very balanced) and how we feel (usually very negative) drives the difference between our intentions (transformation) and our actions (distraction).

The usual suspects, WEF, Deloitte, BCG and the rest, have correctly identified the attributes we need to develop (we analyse these later), and then run into the same practical gap. Identifying capabilities is useful, if directional. But AI is all up in our collective grill, and the world needs something practical, accessible and, frankly, a little less conversation and a little more action, please.

You can't install curiosity or an experimental mindset. Many of us built careers on having the answer ready, and came up as happy little work sausages worshipping at the altar of productivity, climbing the endless ladder of a to-do list. We don't need another set of inspirational but forgettable words on the walls. We need ways to build skills so we can use the tools we've been given, plus enough space and support to develop expertise over time, until it becomes part of how we work and who we are.

To build that bridge, we've gone over five years of data from 3,159 professionals across 25 organisations, all programme participants, and run two 2026 surveys, of our own audience and an independently recruited panel, to test the argument outlined here. The point is a practical route to the capabilities we're all going to need. (You're welcome.)

Since 2021, with UCL support, Uncertainty Experts have been asking people and organisations how they respond to uncertainty. In 2021, six in ten described some sort of procrastination. Delay, distraction, over-analysis, busy work, freeze. By 2023 it was eight in ten, and it has pretty much plateaued since.

Which is awkward, because open-minded curiosity, a beginner's mind and a tolerance for getting it wrong in front of people are exactly what AI adoption needed. By the time ChatGPT earned its reputation for hallucination, a plate-spinning public filed it under another annoying thing to get your head around, somewhere between whether they were too late to start a TikTok and whether they were too early to buy an NFT.

Uncertainty, AI and most interesting people don't fit neatly in boxes. This is a way of noticing what is happening and trying something different. It's a loop: you can go round it three times before lunch, and backwards on a bad Tuesday.

The aim is not to stay in a good box. It is to notice the signal, make room to think, choose the next move and practise until that response is easier to reach for.

That is enough theory for this page. First, let's look at what happens when those capabilities meet an actual working day. We return to the practical route, from practice to capability to identity, at the end.

In our independently recruited comparison panel, three answers sit awkwardly beside each other.

You don't need to build an agent to use AI well. These are not marks out of ten for being clever. But a login doesn't tell you whether someone's work is changing.

In the same panel, 62% had put off an AI tool they believed would help them, and 48% had done so more than once. Roughly three quarters said their organisation encouraged experimentation and made it easy. Permission alone does not explain the hesitation.7

The curve
that didn't fire

The familiar adoption curve is a story about who gets there first and who follows. We're interested in the human bit in between. Someone discovers something useful. Someone else asks the awkward questions. The rest of us get a reason to try it that is more convincing than being told we're falling behind.

That handover is the engine, and it has run for hybrid corn, the fax machine and contactless. With AI, three things went wrong with it at once, and none of them is about the technology. This is our reading of a pattern, not a law. You may recognise it in your own building.

01 The ultra-enthusiasts made joining in look exhausting

I say this as someone very much in the group chat. Keeping your context somewhere permanent, organised and trustworthy is the single most useful change I have made to working with AI. A second brain, in the language of people who have spent quite a lot of time on this. The technique is good. The sales pitch needs work. Try suggesting a second brain to someone whose first has been absolutely fucking pummelled by the last few years, or more usually the last few days, and watch their face.

Then we start trading model names like children comparing whose dad would win a fight. My Astra 6 can beat up your Fable 5. (It probably can. Both will have been replaced by the time you read this, which is rather the point.) The risk isn't that the enthusiasts know too much. It's that their expertise stopped being an invitation and started looking like an entrance exam. Some of us are also knackered. (Not generally how it is advertised.)

02 The useful sceptics were let off

Sceptics do a specific job on every adoption curve, and it is the one nobody thanks them for: they improve the thing. Does this work? What breaks? What happens to the information? What are we giving up? Is this actually better? The technology gets better because it has to get past those questions, which is why the vanguard needs a rearguard. The enthusiasts make the possibilities visible. The sceptics help make them credible.

This time the rearguard was handed an exit. Every previous technology was pitted against another technology, so the sceptic had to argue on features, and the argument made the thing better. This one is pitted against people. "I'd rather talk to a human" is unanswerable, partly because it is right and mostly because nobody wants to be the person in the room arguing against human connection. So the criticism stays and the investigation stops. That is not an instruction to override a sensible objection. Sometimes the tool really is wrong for the job. The loss is the willingness, and the opportunity, to find out together.

03 The middle is buffering

Not adopting, not refusing. Using the approved tool, producing the expected output, getting through the day, and waiting for something to load that never quite does. Nobody has asked them how it is going. From a distance, everything looks reassuringly busy.

We run a live version of Uncertainty Experts called the Leadership Lab. We ask the leaders in the room and their wider teams about the uncertainty facing their organisation, then put the answers next to each other.

That sounds simple until you see the comparison. Across 674 free-text answers in nine organisations, the same question revealed rather different priorities.

The leaders were more likely to be talking about what AI does to the business. Their people were more likely to be talking about what it does to them. Nobody needed to lie for those answers to diverge.

Two strategies are running in one building. Only one of them is on the slides.

This is the moment in a Lab where the room can go quiet. Here is what you said matters most. Here is what your team said. You may all be trying to move the same organisation, but you have not necessarily agreed what is at stake. Before you repeat the instruction more loudly, widen your aperture.

The second gap

We also ask people how well they handle uncertainty themselves, and how well their organisation handles it. A person can be comfortable experimenting while working in a system that makes it difficult. A confident leadership team can underestimate how exposed everyone else feels. Different scores do not tell us which explanation is right. They tell us what to ask about.

The point is not to prove that the top is deluded or the team is resistant. It is to have a conversation the official strategy has not yet had.

The international research cited here puts chronic procrastination at roughly one in five adults across the countries studied. That's the trait version, not the same measure as our participants describing what they do when uncertainty hits. Britain doesn't need to win a procrastination league table for the next argument to matter.9

This is a cultural interpretation.

Look at the vocabulary. The same unproductive afternoon can be "terribly busy" or "works best under pressure" in one telling, and skiving in another. Nobody has ever offered to cure me of working best under pressure.

Our contention is that some respectable versions of avoidance get very little examination. The joke helps. The last-minute rescue helps. Producing something that looks like work helps most of all. Which makes a machine capable of producing endless work-shaped things an interesting arrival.

We have met
the promise before

Every one of these technologies was sold as time given back to you. Each one arrived, and the day got fuller.

None of that is an argument against the technology. It is a reason to design the human part rather than assume it will take care of itself.

You know what you ought to do. You can probably give a very convincing explanation of why it matters. You may even have commissioned a slide about it. And still, nothing moves.

Back to the thought-felt gap, the cornerstone of our work. People can describe sensible, analytical thoughts about uncertainty while reporting something much less comfortable underneath. Their explanation points towards what they would like to do; the feeling often helps explain what they actually do instead.

That matters because a negative response is not the end of the story. It gives us somewhere to work. Fear, Fog and Stasis are the language we use for familiar ways of getting stuck. Action, Agility and Connection are practical alternatives to try, not instructions to cheer up.

There is a use
for the delay

Shin and Grant found a curved relationship between procrastination and creativity, with moderate procrastination associated with greater creativity than either extreme in their studies. That is not permission to leave every deadline until Thursday night. It is a reason not to assume that starting immediately is always the most creative response.10

There is a difference between leaving a problem some room and leaving it for somebody else. Give the pause a boundary. Go for the walk. Have the conversation. Then return to the work and see what has changed. A useful delay has a way back in.

A useful delay has a way back in.

What we did
with the space

Since joining Media Zoo, we've had more room to move from startup fast-talking to deeper thinking: while the wider team helps scale the programme, we've been able to spend more time with the data.

The Decisiveness Crisis helped us develop three lenses for decisions people are actually carrying: Agency, Horizon and Altitude. They turn the findings into questions you can use.

We built a session around them called Ready or Not. Behavioural games, reflection on real decisions, and some protected space in the middle of a working day. The design draws on research into construal level, actively open-minded thinking and perspective-taking. We use before-and-after measures alongside the experience, rather than relying on whether everybody enjoyed themselves.

What kind of leader would you become if you could hold that space?

That is what structured, measured procrastination can make room for.

We started with the lists. The capabilities the future apparently requires: curiosity, adaptability, judgement, exploration, learning how to learn. We agree with the direction. But the blind spot is bigger than how to acquire them.

What happens when the conditions keep changing? When there is no settled version of the job to become good at, and the tools stop feeling like a separate subject called AI, just as new media stopped being new?

Post-certainty is our name for that proposition. Not a prediction that nothing will ever settle, or that planning is pointless. A refusal to build our working lives around the promise that the uncertainty will shortly be over.

Some of the exhaustion comes from holding on until normality returns. When the next model arrives. When the strategy is clearer. When someone tells us what all this means. Then we'll get properly stuck in. Some waiting is sensible. Waiting for the whole picture to stop moving is a poor basis for development.

The next task is not only to acquire a skill. It is to keep adapting it until the capability is something you can draw on when the instructions no longer fit.

Part of the usefulness of James Clear's Atomic Habits is that it connects the person you want to become with something small you can actually do. We need that kind of practicality here: less naming the finished person, more showing them where to start.12

From practice
to identity

This is the positive route through the loop we asked you to recognise earlier. The aim is not to become a different person by Friday. It is to give yourself evidence that another response is possible.

Uncertainty-ready
leadership

The programme, the Leadership Lab, Ready or Not, and the work taking shape around them sit under a banner we're calling Uncertainty-Ready Leadership. The ambition is not a person with all the answers. It is people who can keep learning, thinking and acting together when nobody has them yet.

If that resonates, we'd love to tell you more.