There is a question senior leaders keep raising in private that rarely makes it onto the conference agenda: what happens to our people's thinking once the AI is doing the thinking?

That question has a name — cognitive atrophy — and it was the subject of this SuperCollaboration episode, recorded in San Diego ahead of a closing keynote for a technology and defense conference of around 1,000 attendees. It is not a fringe worry. It is the concern that comes up, again and again, once an organisation moves past the pilot phase and starts deploying AI at genuine scale.

What Cognitive Atrophy Actually Means

Cognitive atrophy describes a decline in critical thinking skills tied to over-reliance on the tools. Not a decline in output — output usually goes up. A decline in the underlying capability: the judgement, the recall, the ability to hold a problem in your head and work it through.

The reason senior leaders are raising it now, rather than three years ago, is scale. A handful of people using an LLM to speed up a first draft is one thing. An entire function defaulting to it for every piece of analysis is another. And the thing about atrophy is that it is invisible while it happens. Nobody files a ticket saying their critical thinking has degraded.

The MIT Study: Three Groups, One Essay, EEG Caps

The research James walks through in the episode comes from MIT. The design is simple enough to explain in a sentence: three groups were given the same essay-writing task, with EEG caps on to measure brain activity. One group wrote using an LLM. One group used search engines. One group used nothing but their own thinking.

The results did not favour the assisted groups.

The Retention Gap Nobody Budgeted For

The brain activity finding is striking on its own. The second finding is the one that should give any leader pause, because it goes to whether the work is being learned at all, not just whether it is being produced.

Think about what that means inside an organisation. A strategy document gets drafted. It gets circulated, approved, filed. And the person whose name is on it cannot tell you, minutes later, what it actually said. That is not a productivity trade-off you can put on a slide and defend. It is a retention failure — and retention is the thing that compounds into expertise.

The Robotic Surgery Parallel

If this pattern feels speculative, it shouldn't. James draws the parallel to robotic surgery, where over-reliance on the machine has been shown to erode surgeons' own skills over time. Same mechanism, higher stakes, and a field that has already had to reckon with it.

That parallel is useful in a boardroom because it moves the conversation off "is AI good or bad" — a conversation nobody wins — and onto a narrower, more answerable question: which capabilities does this organisation need to keep sharp in its people, regardless of what the tool can do?

What Leaders Can Actually Do About It

Here is the part that matters most, and it is the part that gets lost when this research circulates as a headline: none of it is an argument for avoiding AI.

The argument is for deliberateness. Organisations that scale AI without a strategy for protecting critical thinking and decision-making will get the productivity gain and the atrophy together, as a package. Organisations that design for both can take the gain and keep the capability. The difference is not the technology. It is whether anyone at the top decided it mattered.

That is the work: naming which decisions stay human, building the habits that keep judgement in play, and making it explicit rather than leaving it to whatever each team drifts into. It is also, not incidentally, what separates the companies that will still have deep expertise in five years from the ones that quietly traded it away for speed.

Looking for an AI Keynote Speaker in San Diego?

If you're an event organiser with an AI-related theme, this is the material James brings to the stage — the research, the parallels from other fields, and the practical strategies leaders can act on when they get back to the office. He delivers it as an opening or closing keynote for conferences, leadership summits and corporate events, in San Diego and worldwide.

You can explore the SuperCollaboration keynote, see the full range of AI keynote topics, or read the previous episode in this series, recorded in Istanbul.