I can't pair my Bluetooth speakers. But I use AI every day. And I'm starting to wonder what it's doing to my brain.

What AI is doing to the thinking brain

The mechanism is not mysterious. Cognitive offloading is the term researchers use for the process of outsourcing mental work to an external tool, and humans have been doing it for centuries. Writing things down rather than memorising them is cognitive offloading. So is using a calculator, a map, a recipe. The concern with AI is one of scale and speed. Every previous tool automated a specific cognitive task. AI, at least as we are currently using it, can automate the whole chain - the framing of the problem, the searching for information, the synthesis, the conclusion, the writing of the conclusion up. It can, if you let it, do all the thinking and leave you to approve the output.

Neuroscientist Betsy Sparrow's research on what she called the "Google effect" found that people who expect information to be available later remember it less well. The brain, operating with characteristic efficiency, stops bothering to encode things it believes it can retrieve on demand. AI creates an even stronger version of this expectation. Why work through a problem slowly and painfully when the answer is seconds away? The brain learns, quickly, that it doesn't have to.

"The concern is not that AI makes us lazy. The concern is that it is making us lazy in ways we cannot easily see, because the output still looks like thinking."

There is a specific quality to the dopamine hit that AI delivers that makes this particularly sticky. It is immediate. It is responsive. It gives back something that looks like effort when you have put in very little. The brain is not designed to resist that. The brain is designed to seek it out.

The tech bros and their children

It is one of the more clarifying facts about the current moment that the people who built these tools are, with some consistency, keeping them away from their own families. Steve Jobs famously limited his children's screen time. The practice is well-documented among Silicon Valley parents, who send their children to low-tech schools and restrict device use at home with a discipline they do not appear to apply to their professional lives or their valuations.

The question of whether this extends to AI specifically is newer and murkier. Most of the screen-restriction frameworks predate the current generation of generative AI tools. But the underlying logic, that forming minds need friction and boredom and the experience of sitting with a problem until something shifts, applies at least as well and possibly more forcefully. A calculator removes arithmetic. An AI removes the need to have thoughts.

The people who built the dopamine loop have decided, at home, that the dopamine loop is not for their children. This is not a small thing to notice.

Have the rules of leadership gone forever

Every few months someone publishes a piece arguing that AI has made a particular human skill obsolete. Critical thinking. Creativity. Strategy. The argument usually contains a product announcement. The reality, as it tends to be, is more interesting and more inconvenient than either the catastrophists or the boosters are suggesting.

The research on expertise and judgment is fairly clear that the skills most resistant to automation are precisely the ones that look, from the outside, like they are doing the least. The ability to know which problem is worth solving. The ability to read a room and understand what is not being said. The ability to take responsibility for a decision when the situation is genuinely uncertain and all the data points in different directions. An AI can give you a framework for any of these things. It cannot do them for you, because they require something the tool does not have: skin in the game.

What has changed is the table stakes. The cognitive tasks that used to fill leadership - synthesising information, drafting communications, researching markets, modelling scenarios - are being compressed. Leaders who used those tasks to feel busy and call it strategy are going to find themselves exposed. What remains, and what matters more now than it did before, is judgment under uncertainty. The capacity to make a call with insufficient information and live with the consequences. The willingness to have a view. These are not skills that improve with AI assistance. They are skills that atrophy if you stop exercising them.

What you can actually do

The honest answer is that nobody has fully worked this out yet, because the tools are new and the research on their cognitive effects is still catching up. But the neuroscience of skill preservation is reasonably well established, and it suggests a few things.

The brain retains capability through use and loses it through disuse. This is not a metaphor. Neural pathways that are regularly activated are maintained; those that are not are pruned. The implication for AI use is practical rather than puritanical. It is not that you should refuse the tools. It is that you should notice which cognitive muscles you are no longer using and find deliberate ways to use them anyway. Write something by hand occasionally and feel how strange and slow it is. Sit with a problem before typing it into a box. Form an opinion before asking the algorithm what it thinks. Let the discomfort of not immediately knowing be information rather than something to be immediately resolved.

The people who will navigate this well are not the ones who avoid AI out of principle. They are the ones who use it with enough self-awareness to know when they are thinking and when they are just watching something else think for them. That distinction is going to matter quite a lot in the next ten years, and it requires exactly the kind of attention that the dopamine loop is designed to make difficult.

I cannot pair my Bluetooth speakers. But I know when I am doing my own thinking. At the moment, that feels like the more important skill to hold on to.