Are we quietly outsourcing our ability to think?

You catch yourself doing it before you have even decided what you think. A decision lands in your inbox, something that two years ago you would have sat with, and your hand is already opening the AI window. You reach for the answer before you have formed your own view. And then, over a few weeks, you notice your team doing exactly the same thing in meetings, pulling up a generated summary instead of saying what they believe. You are all moving faster than ever. And quietly, in a place you can feel but cannot quite name, you worry you are all becoming less able to actually think.

Here is the answer to the question you are circling: yes, that worry is real, and no, the danger was never the robots. The risk that matters at the team scale is that capable people de-skill their own judgement without noticing, because the tool is so good and the moment of reaching for it costs so little. The good news is that this is a habit, not a fate. It can be made conscious. That is the whole game.

Am I actually losing my ability to think, or just moving faster?

Both, and that is the trap. The speed is real and so is the cost, and they arrive together so you only register the speed. Microsoft Research and Carnegie Mellon studied this directly in 2025, gathering 936 real first-hand accounts of generative AI use from 319 knowledge workers. The finding is sharp enough to pin to your wall: the higher a person's confidence in the AI, the less critical thinking they reported doing. The higher their confidence in their own ability, the more critical thinking they reported. Reliance and thinking moved in opposite directions. For basic recall and comprehension tasks, around 72% of participants reported putting in less mental effort when they used the tool.

Now sit with the mechanism, because it is the part that explains your team. The study named the conditions under which people skip their own thinking, and two of them are simply time pressure and lack of awareness. That is your inbox at 4pm. You are not lazy and neither is your team. You are responding rationally to load, the tool is right there, and the offload happens below the level of choice.

What does the loss actually look like inside the head?

There is a physical version of this, and it is worth knowing. A 2025 MIT Media Lab study put 54 people in an EEG cap (a headset that records the brain's electrical activity through the scalp) and had them write essays, some with an AI assistant, some with a search engine, some with only their own minds. The AI group showed the weakest neural connectivity of the three. The researchers put it plainly: brain connectivity scaled down in relation to the tool. The AI users also reported the lowest sense of owning what they had written, and many struggled to quote a single sentence from the essay they had just finished. The authors gave it a name that should stop any leader cold: cognitive debt (a lasting thinking deficit that builds up when you keep offloading the work to a machine). A deficit that did not clear when the tool was taken away. The study is a preprint (research shared publicly before formal peer review), so hold it as a strong signal rather than settled law; the direction is hard to ignore.

The effect is not confined to a lab. A 2025 study in the journal Societies, surveying 666 people across a broad span of ages, found a significant negative correlation between frequent AI use and critical thinking, with the link running through cognitive offloading, the habit of letting a tool do the reasoning. The youngest participants, who leaned on the tools most, scored the lowest of all. Two very different methods, the lab and the survey, point the same way.

This is the failure mode worth fearing. Not incompetence, which announces itself. Pseudo-competence, which stays hidden. People who feel sophisticated while quietly losing the ability to spot a fundamental flaw, because the flaw-spotting muscle is the one being rested.

Look at that Gallup gap. Nearly half your people are using AI on the job, and barely a third work somewhere that brought it in on purpose. The thinking habits of your organisation are being rewired by accident, in the space between official policy and actual behaviour. Shadow use is setting the shape of how your best people reason. That is the thing to take seriously.

How do we use AI and keep our judgement sharp?

This is a question of conditions and sequence, and it sits firmly in the human layer. Deloitte's 2026 State of AI in the Enterprise report, drawn from 3,235 leaders across 24 countries, found that the single biggest barrier to putting AI into real workflows is no longer the model or the infrastructure; it is insufficient worker skills. Read that again. The leaders closest to the spend are telling you the constraint sits with the humans, not the machine. The bottleneck is no longer the technology. It is whether your people still bring their own mind to the table before the machine speaks.

So you protect two things: the sequence of thinking, and the state in which thinking happens. The sequence is a discipline anyone can run. The state is more physical than most leaders admit. Considered judgement has a measurable substrate; a 2022 meta-analysis (a study that pools the results of many earlier studies) in the neuroscience journal Cortex found a small but consistent positive link between higher heart rate variability, the natural beat-to-beat variation that reflects a calm, regulated nervous system, and executive functioning, the top-down brain control behind self-regulation and good decisions. And decision quality decays under load: the well-known "hungry judge" parole study found favourable rulings starting near 65% straight after a food break and sliding toward zero before the next one, then resetting once the judges had eaten and rested. Treat that as illustrative rather than ironclad, since the replication debate is live, but every leader recognises the felt truth of it. Depleted people reach for the easy answer. AI is the easiest answer ever built. Guard the state, and you guard the judgement.

1. Form your view first. Write a line on what you think before you open the tool. Even one sentence. This is what the research links to sharper critical thinking, not weaker.
2. Use AI to challenge, not to author. Bring it in as the second voice. Ask it to argue against you, to find the hole, to name what you have missed.
3. Make people quote their own reasoning. In reviews, ask the person to defend the decision in their own words. If they cannot, the AI thought and they did not.
4. Decide the high-stakes calls when rested. Protect the conditions of judgement the way you protect a budget. The state is part of the work.
5. Name the offload out loud. Make "I reached for it before I'd thought" a normal, safe thing to say on your team. Awareness is the antidote the studies point straight toward.

The leaders who win the next decade are the ones who upgrade themselves first. That means choosing, on purpose, to remain the kind of person and the kind of team that can still form a view, hold a position, and recognise a flaw, with the machine as a partner in that work rather than a replacement for the mind doing it. Speed you already have. The aim now is to keep the judgement that makes the speed worth having, and to use these tools as consciously as you would want your sharpest people to wield them.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.