One of the most difficult things to do is to create an effective mechanical way of measuring that student has learned the subject matter.
Naturally for some subjects it is more straightforward than others, but the problem is still challenging on balance.
Early childhood development requires a lot of psychological growth, in addition to knowledge growth. Abstract thought is not always acquired at different rates, and recognizing its emergence isn't very deterministic.
A solely software-based system also requires that the user provide honest input. It's hard to diagnose a lying patient.
Some parts intuition, some parts adaptability.
As the ability of software to handle these tasks improves, it will surely become a larger part of the equation. I don't think it will replace live humans in my lifetime though.
Naturally for some subjects it is more straightforward than others, but the problem is still challenging on balance.
Early childhood development requires a lot of psychological growth, in addition to knowledge growth. Abstract thought is not always acquired at different rates, and recognizing its emergence isn't very deterministic.
A solely software-based system also requires that the user provide honest input. It's hard to diagnose a lying patient.
Some parts intuition, some parts adaptability.
As the ability of software to handle these tasks improves, it will surely become a larger part of the equation. I don't think it will replace live humans in my lifetime though.