Is Our Children Learning? Apparently Not.

So much going on these days, we’re throwing the regular Friday roundup format out the window. Recommended reads below, but first

Let’s begin with the appalling news that “American Kids Are Getting Even Worse at Reading.” It’s hard to overstate the catastrophe of this trend, which began for the bottom half of test-takers pre-Covid, worsened with school closures, and shows no sign of reversing. After decades of extensive research, efforts at reform, increases in funding, deployments of technology, we just hit new lows. We are doing a worse job teaching kids to read than in 1990, and inequality between top and bottom is only getting wider.

At no point during this slide, so far as I can tell, have policymakers or the education establishment looked up from the hole they seem to be digging and thought, maybe we should stop? No, the fact that new-age techniques seem only to make things worse yields only calls for yet more newer-age techniques. The total failure of technology to deliver progress leads to the obvious conclusion that we need to bring even more technology into the classroom even faster. In parallel, as therapeutic trends place ever greater emphasis on social and emotional health that only continues deteriorating, someone somewhere has apparently dictated that the answer must be to emphasize it more and coddle better.

Now we seem poised to double down yet again, this time on AI as “the greatest education innovation ever.” That’s what Tyler Cowen called it, at a PR rally (sorry, workshop on AI economics) hosted by OpenAI that I attended earlier in the week. More on that in a moment, but the idea of AI as an ideal “teacher” continues to pop up, and strikes me as an especially bizarre conceit at the intersection of everything we are getting wrong about both education and AI. At Persuasion this week, Yascha Mounk similarly enthused that, “The ways in which AI can help you learn a language are incredible. You can have a conversation in the language you’re practicing with the world’s cheapest and most patient tutor.” Yes, yes, there are “challenges of design, pedagogy and user experience,” but “somebody will almost certainly fix these problems.”

Look, I don’t doubt that the Tyler Cowens and Yascha Mounks of the world can make great use of AI in their own learning. But educating Tyler and Yascha isn’t the challenge. And for the typical learner, let alone somebody who really struggles, the lack of technological tools is not the obstacle. Engagement, confidence, pedagogy, relationship… the things an educator actually has to get right have nothing to do with the “power of the model,” or its UX design, and no amount of coding will change that, just as none of the billions of dollars poured into education technology over the past 30 years appears to have achieved anything at all.