Last week, I told you about the one part of oral assessment I still won’t hand off to AI. This week, from a completely different direction, someone else landed on almost the same instinct — without knowing anything about what I’ve been building.
I asked her one question: what’s the single science-of-learning principle every teacher should hold onto in the age of AI? Her answer: be as metacognitive about the AI as you are about your own thinking. Simple to say. Genuinely hard to picture in an actual classroom — which is most of what this conversation is about.
That guest is Dr. Tina Grotzer, who runs the Next Level Lab at Harvard’s Graduate School of Education, studying causal reasoning — how people actually trace cause and effect, versus the neat, linear version we usually assume.
That’s three separate people now — me, an MIT researcher, and a Harvard researcher — landing in the same place, independently. I don’t think that’s a coincidence anymore.
Key Moments
1:33 – What “causal reasoning” actually means (her research, in plain terms)
5:10 – The sinking-and-floating experiment that breaks kids’ assumptions
10:40 – Introducing the Next Level Lab: will AI make us smarter or stupider?
17:13 – Her core thesis: why metacognition is what makes AI a partner instead of a crutch
19:32 – The Russell Hulse story — a Nobel Prize won by refusing to ignore an anomaly
21:11 – The “forklift to the weight room” metaphor, and her fear of AI as a crutch
27:06 – You lay out AI Engines, Analog Roots, and the Mastery Viva live — and she reacts in real time
34:11 – The one cognitive skill she says AI will be the last to replicate: empathy
40:00 – Her closing fear: a coming “bifurcation” between kids taught to think and kids who aren’t
