I’m always surprised EdTech hasn’t done more to transform education over the last thirty years.

On a very simple level, education is: tell them stuff, get them to do stuff, assess what they learned from the stuff they do, and repeat. The debate around discovery learning versus explicit instruction is, at its crudest level, just about the order of the steps. It’s strange that in nearly three decades, technology has been unable to execute these processes dramatically better than humans, despite billions of investment. It’s especially puzzling given that in the same period, so many other areas of human endeavour have radically changed and mostly improved through technology.

So either human-led education is already close to optimal, or there are missing ingredients that computers simply can’t provide.

That’s why I’m skeptical about claims that AI will be transformative. It may improve the telling, the doing, and the assessing. But I’m not convinced AI can do these so much better than existing technology can today. And no one seems to be seriously searching for the missing ingredients.

I hope I’m wrong. Education remains an expensive, highly skilled labour-intensive process, and that limits who can benefit from it being done really well.