This is Part 2 of 5 in the Anti AI sentiment arc. If you’d like to see where we’ve just been, the post before this one is What the Room Was Holding.
A quick orientation, if you’re new here: this follows directly from the evening with old friends where AI came up unexpectedly and revealed opinions I hadn’t known were sitting there. I’m not going to re-tell that night here — you can read the full version. What I want to get into is the thing that stayed with me after the laughing stopped.
Two hours after Oliver asked his first question about AI, I sat back and listened to a case being built against it — confident, fast, and complete. What struck me afterward wasn’t the case itself. It was that nobody had ever taught the people building it how to notice they were building it at all.
The juxtaposition underneath this one isn’t AI-literacy versus AI-illiteracy. It’s older and sharper than that: the capacity to examine your own convictions, versus never once having been shown that convictions need examining in the first place. Oliver and my son didn’t arrive at their position through reasoning I could follow or push back against — there was nothing to push against, because the position had never been tested from the inside. It had simply accumulated, the way a feed accumulates, one confirming data point at a time, until it stopped feeling like an opinion and started feeling like something they’d always known.
The missing skill has a name — metacognition, thinking about your own thinking. The term was introduced in the 1970s by the psychologist John Flavell, and the approach that builds it is direct instruction: showing people the mechanics of thinking rather than hoping they pick it up. One recent study of faculty at three Harvard professional schools found something that lines up almost exactly with what I watched that night: 54 percent of the faculty surveyed said they explicitly taught critical thinking in their courses. When researchers talked to the students, the picture was different — critical thinking was happening implicitly, if at all, through discussion and debate, with no direct instruction in the actual mechanics of it. Students generally couldn’t recall or define metacognition when asked. The skill everyone assumes gets absorbed somewhere along the way — usually isn’t. Not explicitly. Not on purpose. Critical thinking, when it’s actually turned on yourself rather than someone else’s argument, is a form of metacognition, and most people are never handed the instructions for it.
I don’t think anyone formally taught me to check a belief before defending it either — but forty-one and a half years of being the person called in when something had broken taught it to me anyway, by necessity, because you can’t fix a system without first asking why it actually failed, not why you assumed it had. Watching Oliver and my son defend their wall, I understood that instinct was never handed to either of them. Nobody ever sat them down and said: before you decide what you believe, ask where it came from.
I got that training by accident. Decades of it, one broken system at a time, until asking “why did this actually fail” stopped being a technique and just became how I thought. Oliver and my son haven’t had forty-one years yet. They’ve had a handful of years and an algorithm that never once asked them to check its work, because checking its work was never the algorithm’s job. It was supposed to be someone else’s — a teacher’s, a parent’s, mine, maybe, if I’d known to do it sooner.
That’s the gap.
Not a gap in intelligence, or in caring, or in capability. A gap in ever being handed the one question that would have let them catch it happening: where did this belief actually come from, and have I ever tested it against anything other than more of the same?
I don’t have a tidy fix for that gap in one blog post. But naming it plainly feels like the right place to start — because you can’t teach yourself a skill you don’t know is missing, any more than Oliver and my son could examine a wall they didn’t know they’d built.
Sandi is a Melbourne-based problem-solver, crisis-averter, and translator of the technical into the human. She spent decades being the person everyone called when something was broken, confusing, or just needed explaining properly — earning a reputation that preceded her wherever she went. Now she’s channelling that same instinct into AI: making it accessible, practical, and genuinely useful for people who think it isn’t for them.
