AI Redefining Human Cognition

I want to be careful here, because 'AI is redefining human cognition' is the sort of headline that sounds profound until you ask what it actually means, at which point it tends to dissolve.

And yet something is clearly happening. People are genuinely using these tools differently to how they used Google, differently to how they used calculators. The cognitive relationship is different.

The cognitive relationship with the television was also 'different'. We did not convene seminars about it redefining human cognition. We just watched more of it and became slightly worse at reading.

That's actually not a bad parallel, but it cuts against you. We probably should have convened those seminars. The lesson is worth taking seriously precisely because we didn't do it then.

Fine. I'll concede that. The question I'd rather be asking is whether AI is augmenting cognition or replacing the parts of it we find inconvenient.

The lesson frames it as augmentation — 'AI-augmented memory', 'collaborative problem-solving' — but I think that framing is doing quite a lot of optimistic work.

It is doing all of the work. 'Collaborative problem-solving' is a touching way to describe asking a machine to solve the problem while you watch.

Except that isn't always what's happening. A researcher using GPT-4o to stress-test an argument is genuinely thinking harder, not less. The tool is creating friction in the right places.

That researcher exists. She is not the median user. The median user is asking it to write their email and feeling vaguely clever for having done so.

You can't build a framework around the median misuse, though. We don't evaluate literacy by looking at people who only read cereal boxes.

We do, rather, when we're trying to design educational interventions. Which is what this lesson is.

That's a fair point. And the lesson does try to address it — there's a whole objective around formulating strategies for human-AI cognitive collaboration, which at least acknowledges you have to be intentional about it.

The ethical implications section interests me more. Because the uncomfortable version of this conversation is that if AI is genuinely reshaping memory retention and critical thinking, we should probably discuss who benefits from that reshaping and who designed the systems doing it.

Now you're asking the question the lesson gestures at but doesn't quite land on. There's a difference between 'AI changes how we think' as a neutral observation and 'specific AI systems, built by specific companies, are changing how we think' as something requiring scrutiny.

Yes. And I suspect most people will complete this lesson feeling broadly optimistic about augmented cognition, which may or may not be the appropriate response.