AI is a Tool - not a replacement for your brain

AI really should not be used instead of your brain, says Hari, especially when it comes to learning and self-improvement…

Image of toolbox, a versus sign, image of a brain

I’ve been thinking more and more about the attendees on a course of mine who came along, gave me 3 and a half hours of their virtual attention, and then used AI to do the final exercise.

Lovely people all, and it’s certainly true I never said “don’t use AI for this exercise” but that was mostly because it never occurred to me anyone would. (More fool me).

In fairness, one of the consistent pieces of feedback I get about the two big group exercises for this particular course is that “there wasn’t enough time.” These two exercises are usually about 10-15 minutes to improve some badly designed PowerPoint slides and 15-20 minutes to do a bit of analysis on data provided. And yes, that’s not enough time.

But this is a training course - if we replicated reality accurately then you very rarely have only 15 minutes to improve PowerPoint slides. Even in the highest pressure of circumstances you’ll usually have an hour or more to consider and work on them. And no reasonable boss would only give you 20 minutes to analyse some spreadsheets.

But this is, as I’ve said, a half-day training course. We don’t have several hours for each activity, they need to be contained within the course and a little bit of time pressure can produce some brilliant results. It is, after all, a chance to practise newly learned skills.

So I do understand the instinct - you’re low on time, the pressure is on, and the AI prompt box is right there…

However, general principles of learning say that you learn and retain much more by doing rather than listening. So it stands to reason that if instead of doing the work you are choosing to write a prompt to make AI do the work, you’re going to learn even less.

There’s a few different versions of it about, but the Learning Pyramid generally looks like this:

People generally remember: 10% of what they read, 20% of what they hear, 30% of what they see, 50% of what they see and hear, 70% of what they say and write, 90% of what they do

That means that if we use AI to cover the “do” section, we’re instantly reducing the memorability of the material learned by 20% because we’re not actually doing the work. We’re just writing a prompt for something else to do the work.

Now that’s not to say there is no place for AI in training. I was genuinely inspired by a demo I saw earlier this year using AI to help people practise having difficult conversations with virtual team members. I’m been tempted to try and create a similar bot to enable students to practise verbal mirroring and listening for subtext. Just haven’t had the time to play with it yet.

But if we let ourselves out-source our cognitive abilities, if we spend hours learning about analysis and storytelling but when it comes time to practise we hand it over to AI, well, at the very least I think that is somewhat missing the point.

Previous
Previous

September 2025 - round-up

Next
Next

Apple Keynote 2025 talks design, photography, hardware - but not AI…