What struck me most is that we’re not really moving from “bad system → better system”, but from one form of control to another. Some AI implementations share a similar flaw: they optimize for compliance, not for actual learning.
The opportunity with AI isn’t automation of teaching — it’s finally breaking the industrial model of education. But that only happens if we design systems that understand the learner, not just generate answers. Otherwise we just end up with very articulate, scalable mediocrity.
For me, the key question is:
Do we use AI to enforce the system more efficiently, or to redesign it around human development?
I agree to an extent, but any system will always operate with forms of control, after all, even self-governance is a form of control.
Respectfully, I think it's an unhelpful dichotomy to see education as either "teacher-centred" or "learner-centred" (which I think is what you are suggesting by the 'industrial model', but please do clarify if I've misunderstood your meaning).
What do you see as the outcomes of a system redesigned around human development; or to put it another way, what does the alternative to 'very articulate, scalable mediocrity' look like to you?
That’s a fair push :) and I agree, “control” itself isn’t the problem. Every system has constraints. The real question is what is being optimized for within those constraints.
And you’re right to challenge the dichotomy — I don’t think it’s simply teacher- vs learner-centred. What I meant by the “industrial model” is a system optimized for standardization: same pace, same outputs, same measures of success — regardless of individual context.
So for me, a system designed around human development would look like:
Progress measured by capability growth, not just content coverage
AI used to adapt pathways, not just deliver answers faster
Teachers acting more as interpreters of learning signals than content transmitters
And importantly — space for friction, struggle, and thinking, not just efficiency
Because “scalable mediocrity” emerges when we remove too much of that friction in the name of productivity.
The alternative isn’t chaos or full personalization of everything — it’s systems that are structured, but responsive. Where AI helps us see the learner more clearly, not flatten them into averages.
Great advice for education leaders, and edtech founders shouldn't confuse obsession with the problem for obsession with the product.
Just because you can build another feature, doesn't mean you should!
I really like the tension you highlight here.
What struck me most is that we’re not really moving from “bad system → better system”, but from one form of control to another. Some AI implementations share a similar flaw: they optimize for compliance, not for actual learning.
The opportunity with AI isn’t automation of teaching — it’s finally breaking the industrial model of education. But that only happens if we design systems that understand the learner, not just generate answers. Otherwise we just end up with very articulate, scalable mediocrity.
For me, the key question is:
Do we use AI to enforce the system more efficiently, or to redesign it around human development?
Because both paths are already happening.
I agree to an extent, but any system will always operate with forms of control, after all, even self-governance is a form of control.
Respectfully, I think it's an unhelpful dichotomy to see education as either "teacher-centred" or "learner-centred" (which I think is what you are suggesting by the 'industrial model', but please do clarify if I've misunderstood your meaning).
What do you see as the outcomes of a system redesigned around human development; or to put it another way, what does the alternative to 'very articulate, scalable mediocrity' look like to you?
That’s a fair push :) and I agree, “control” itself isn’t the problem. Every system has constraints. The real question is what is being optimized for within those constraints.
And you’re right to challenge the dichotomy — I don’t think it’s simply teacher- vs learner-centred. What I meant by the “industrial model” is a system optimized for standardization: same pace, same outputs, same measures of success — regardless of individual context.
So for me, a system designed around human development would look like:
Progress measured by capability growth, not just content coverage
AI used to adapt pathways, not just deliver answers faster
Teachers acting more as interpreters of learning signals than content transmitters
And importantly — space for friction, struggle, and thinking, not just efficiency
Because “scalable mediocrity” emerges when we remove too much of that friction in the name of productivity.
The alternative isn’t chaos or full personalization of everything — it’s systems that are structured, but responsive. Where AI helps us see the learner more clearly, not flatten them into averages.