AI might accelerate processes, but it can’t sit with ambiguity

One of the most quietly dangerous phrases in modern workplaces is: “Can we just get a quick answer?”

It is dangerous because what people often actually mean is: “Can somebody make the discomfort of uncertainty disappear?”

Modern organisations are obsessed with speed – faster responses, faster systems, faster delivery, faster decisions. Everyone is “circling back”, “touching base” and escalating things marked urgent that absolutely were not urgent five minutes earlier. Meanwhile, half the workforce is running on caffeine, cortisol and Microsoft Teams notifications, so naturally we’ve collectively decided AI will fix this.

To be clear, AI is incredibly useful. I often use it myself. It can reduce administrative friction, process huge amounts of information quickly and automate repetitive tasks that human beings probably should not have been doing manually in the first place. But there is a significant difference between accelerating processes and exercising judgement, and organisations are increasingly starting to confuse the two.

AI performs brilliantly in structured environments: clear rules, defined outputs, historical patterns and predictable systems. The trouble is that human organisations are almost never like that. Most genuinely difficult workplace decisions happen inside ambiguity – not spreadsheet ambiguity, but human ambiguity.

That is the kind where nobody has the full picture yet, where the data technically says one thing but your operational instincts are screaming another, where a safeguarding concern first appears as “something feels slightly off”, or where culture problems emerge through strange tension in meetings long before they appear in engagement surveys. Risk rarely arrives neatly labelled. More often, it develops quietly through accumulated workarounds created by exhausted people trying to survive impossible workloads.

A lot of governance work – real governance work – is essentially professional pattern recognition under conditions of uncertainty. That is also why I am increasingly unconvinced by the notion that the future belongs entirely to the people who can produce answers fastest. Sometimes the most valuable person in the room is the one saying: “Hang on. I don’t think we fully understand what’s happening yet.”

Unfortunately, many workplaces still reward the opposite. Confidence is often treated as competence, while fast responses get mistaken for good judgement. Reflective people can be perceived as hesitant simply because they insist on sitting with complexity for longer than is socially comfortable. Corporate culture still has a strange tendency to reward performance signalling over actual thinking, and AI may accidentally make some of this worse rather than better.

Because AI tends to mirror the logic of the systems around it, if an organisation already prioritises speed over reflection, visibility over substance and certainty over nuance, then introducing AI into that environment may simply accelerate existing dysfunction more efficiently – which is not quite the futuristic utopia everyone put in the PowerPoint.

I also find this conversation intersects interestingly with neurodivergence. Many neurodivergent professionals, particularly those used to navigating unpredictable or cognitively demanding environments, often develop strong systems-thinking and pattern-recognition abilities. Sometimes, we identify operational tensions long before those tensions become formally visible to everybody else.

At the same time, many workplaces still assess professionalism through communication style rather than decision quality: who sounds polished, who speaks confidently in meetings, who performs calmness convincingly enough, who understands the unwritten social choreography of corporate environments. Those things are not always the same as good judgement.

And in increasingly AI-enabled workplaces, that distinction matters enormously because eventually somebody still has to sit in the room where the information is incomplete, the politics are unspoken, the risks are emerging, everybody is uncomfortable and there is no clean answer yet.

AI is excellent at generating outputs. It is still remarkably bad at sitting with uncertainty without hallucinating confidence. And frankly, quite a lot of humans are bad at that too.

I suspect ambiguity tolerance – the ability to remain thoughtful, reflective and operationally calm without forcing premature certainty – is quietly becoming one of the most valuable workplace skills of the next decade. The irony is that many organisations still do not recruit, reward or promote for it nearly as much as they should.