Midweek Quick Take: what about AI learning?
A reader wonders what I think of Alpha School
Relieved to find the world still here when I woke up this morning. So on the basis that the End Times might not be Happening Right Now after all (phew, and let’s hope it stays that way) a quick post to share something that’s been on my mind about AI and learning.
A reader messaged me to ask what I think of the Alpha school, where children reportedly get amazing results by spending two hours a day being tutored by AI and the rest doing other, more active stuff.
Doesn’t this argue against the need to re-humanise learning, he wonders? Couldn’t we perhaps use AI to replace annoying, inefficient human teachers?
Ok, so: I don’t know a huge amount about this school, but from what I’ve seen it seems on the face of it not crazy to try and tilt kids’ experience back toward social formation. Especially if the parents aspire for the kids thus formed to become tech-elite adults, rather than factory workers. Even so, I reject out of hand the implicit assumption that knowledge is an inert substance that can be delivered by robots. This is a predictable category error for tech people to make, but a category error nonetheless.
Knowledge can be codified to an extent, but making it your own always requires movement toward knowledge by the learner. Classically, that happens in relationship. To illustrate: in our home, we use Duolingo to support my daughter’s language learning, and there are clearly some benefits to digital tutors of this kind. But from my observation there’s an additional stage to learning, where the material must be metabolised and then applied in a human-to-human context. By definition the robot can’t supply that.
In our case, for example, metabolism doesn’t really happen until our summer visit to family friends in Italy, where she gets to practice what she’s learned. From observation, an impressive level of Duolingo proficiency doesn’t map straightforwardly into the same level of proficiency in the wild, talking to other humans. But this isn’t to say AI-assisted language learning has zero effect. She clearly has more of an intuitive grasp of the structure of Italian now, than she would have done without the Duolingo practice.
Is this overall any better than the same amount of time spent in industrial-type classrooms, sitting through Italian lessons taught by a human? Hard to say. Certainly, I don’t know that classroom-based Italian lessons at school over the same period would have delivered more understanding. My hunch, though, is that either would pale into insignificance next to the same amount of time spent learning Italian one-to-one from a human tutor. I loved Will Orr-Ewing’s recent post on this topic:
All of this is to say I think it’s a mistake to infer, from the results achieved by a small cohort of carefully-parented offspring of wealthy, high-IQ tech-world parents, that AI education can be beneficially scaled to everyone. At best it suggests there might be smart ways it can be used as an extension to existing systems, if employed judiciously alongside the indispensable relational component of learning.
My gut feel, though, is that if we assume that knowledge is inert, and can be delivered by robots including to the less able and those unsupported by good home environments, the result will be no better, and perhaps worse, than even the current obviously sometimes sub-optimal industrial, classroom style of teaching. As for how such education contributes to overall formation of persons, even elite ones, it’s probably best to withhold judgement on the Alpha education model until we can see its adult fruit.
What do you think? Are you rabidly anti-AI in all respects? Suspicious of it in learning? Cautiously supportive as a tool? All in on replacing teachers? Tell me what you think in the comments.
Michael Oakeshott:
"Education is not acquiring a stock of ready-made ideas, images, sentiments, beliefs and so forth; it is learning to participate in the conversation of mankind"
He's right, of course.
Last April my husband and I attended on AI and the Future of Education at the Perimeter Institute for Theoretical Physics and were expecting to hear a lot of hype about how it would revolutionize education. Instead, to our great surprise, the speaker (Derek Muller) emphasized that 1. AI is used to render learning easier, thus bypassing effortful engagement and fundamental "system 1" thinking, 2. the real innovation lies not in using yet another technology, but people, because learning happens in the context of relationships.
Learning = effort + relationship / information
To learn something, we need to use cognitive effort, and we need to do it through human relationships; and since we can only absorb so much information at once, we need to present just enough information to challenge students but not to overwhelm them. Then repeat until mastery, and keep doing it.
You can read our essay reporting on this talk here: Learning Fast, and Slow: Why AI will not revolutionize education https://schooloftheunconformed.substack.com/p/learning-fast-and-slow-why-ai-will
Over the last two decades I've repeatedly observed that what propelled my children's learning was not the subject matter per se, or who well the curriclum was designed, but whether they connected with and admired their teachers. My daughter ended up studying French in universtiy because she was deeply inspired by her French teacher; my son pursued medieval studies because of the incredible passion his high school teacher brought to the class.
Passion for learning may ignite through subject matter alone, but it is the human teachers that bring it to life who act as catalysts.