AI Isn’t Killing Critical Thinking. Your Meeting Culture Did That First.

There is a particular kind of headline doing the rounds at the moment, and it goes something like this: AI is making us stupid. The evidence underneath it is more serious than the headline suggests. A 2025 study by Michael Gerlich, surveying 666 people, found that the heavier someone’s reliance on AI tools, the weaker their measured critical thinking, an effect explained largely by what researchers call cognitive offloading: the habit of handing our thinking to an external system. The 2026 International AI Safety Report points the same way. It notes one finding that should stop any leader mid-scroll. In a study of clinicians who had spent three months working alongside AI assistance, their unaided ability to detect tumours dropped by around six percent. Use it, lose it.

It is tempting to read all of this as a story about technology. It is really a story about conditions, and those conditions were deteriorating long before anyone typed a prompt.

The erosion didn’t start with the chatbot

The cognitive scientist Maryanne Wolf has spent years describing what she calls the changing reading brain: the way a culture of skimming, scrolling and constant interruption quietly weakens the deep, reflective attention that real thinking depends on. That shift predates generative AI by a decade. The open-plan office that never goes quiet, the calendar with no white space, the meeting culture that rewards the fast answer over the considered one, all of these were already training people out of deep thought. AI did not introduce the problem. It poured accelerant on it.

That distinction changes what the solution looks like. If you believe AI is the cause, your instinct is to restrict the tool, with usage policies, bans and nervous memos. But if the real issue is that your organisation offers nowhere to think, restricting AI just leaves people with the same impoverished conditions and one fewer crutch. The thinking still does not happen.

Cognitive offloading is not the enemy

Offloading is not inherently bad, and the panic tends to rush past that. We have always offloaded. Writing is offloading. A calculator is offloading. The research is clear that delegating routine cognitive load can free up capacity for harder, more reflective work, but only if that freed capacity is actually spent on harder, more reflective work. The real danger is narrower. It is offloading the reasoning itself, the analysis, the weighing, the slow business of staying with a problem, and then never reclaiming the space it opens up.

Gerlich’s study found one factor that buffered people against the decline: education, or more precisely, the learned habit of reflection. People who had been trained to pause and interrogate held onto their critical thinking regardless of how much AI they used. The skill that protects you is not avoiding the tool. It is the discipline of deciding what to think through for yourself.

That discipline is exactly what most workplaces have stopped cultivating.

Thinking is a condition you build, not an instruction you give

In the work I do with leadership teams, two of the conditions we design for deliberately are what I call the Silent Space and the Thinking Space, and they are not the same thing. Silence is the absence of noise: the meeting that does not fill every pause, the inbox that is not checked mid-conversation. Thinking is what silence makes possible: the slow, sometimes uncomfortable work of staying with a problem before resolving it.

Most organisations have removed both. We have built environments in which there is no moment in the day when a person is expected, or even permitted, to think without an input arriving. And then we wonder why the ideas are shallow and the reasoning is thin.

You cannot instruct your way out of this. Telling people to think more critically while they sit in twelve hours of back-to-back meetings is like telling someone to relax while you stand on their foot. The leader’s job is not to demand better thinking. It is to build the conditions in which thinking can happen, and then to protect those conditions, because almost every pressure in modern working life is conspiring to fill them.

What this looks like in practice

• Protect a pause in the room. When a hard question lands, resist the urge to fill the silence. Let people think before they speak. The first answer is rarely the best one. It is just the fastest.
• Separate generating from judging. Use AI freely to generate options, drafts and first passes. Then close the laptop and do the judging yourselves. The reasoning is the part you cannot afford to outsource, because it is the part that atrophies.
• Build reflection in as structure, not virtue. Do not hope reflection happens. Schedule it. A standing pause before a decision is locked, where the only question is whether you are solving the right problem, is worth more than another hour of generating options.
• Model it from the top. If the most senior person in the room treats their own thinking as something worth protecting, everyone else gets permission to do the same.

The teams that pull ahead over the next few years will not be the ones that use AI the least, or the most. They will be the ones whose leaders rebuilt the conditions for thinking that modern work had quietly stripped out, and who use AI to do more of the work worth doing rather than to stop thinking altogether.

So the real task for leaders is not policing the tool. It is rebuilding the room around it.

---

The Strategic Solutions Lab is a facilitated engagement for leadership teams working through complex challenges, built around structured reflection rather than faster answers. Start a conversation.