Enjoyed this piece, Andrew! As a fellow parent who uses AI to supplement/elevate my children's education, the studies you laid out are compelling and it's a real shame that the wrong studies are being used to draw the wrong conclusions about what is suitable as an educational tool for children.
A few thoughts:
1. The term "ban" is yet another thing that the media has gotten wrong. Norway hasn't actually banned AI for children (which implies legislation and comes off as more draconian). Rather, the Directorate of Education has provided a national recommendation against the use of AI.
In practice, it may be that many schools would follow the recommendations set out, but to me it's not entirely the same thing as a ban that has been passed into law. And parents who are so inclined (like you and me) can still use AI at home to teach their children.
Whilst one could argue that having such a national recommendation in place can taint the image of AI in general and plausibly lead to fewer parents using it for education even at home, using technology judiciously in education to obtain better results with minimal downsides/side effects (eg. cognitive offloading) has always been a question regardless of the era and not a problem specific to AI, so it looks more to be a user-related matter.
2. "Norway forecloses the one configuration the evidence actually supports: a tutor that withholds the answer and walks a student toward it." - my interpretation of their recommendation against AI for specific ages was that this role of withholding the answer and walking the students towards it is and should still be the teacher's responsibility so it's not necessarily that they've taken that option away entirely.
Thank you, Jane - these are sharp, and you’re right on the terminology. A “recommendation” isn’t a “ban,” and I should’ve been more careful with the frame. Why I think it still matters: a national recommendation carries a chilling effect even without the force of law - it shapes what teachers feel licensed to try, which is close to your point about it tainting AI’s image even for home use.
On (2), this is the crux and it's a great pushback. You’re right that withholding the answer and walking a student toward it is the teacher’s job and always has been. My worry is scale - a good teacher does this beautifully one kid at a time, but can’t for 25 at once. That Socratic, answer-withholding role is exactly what a well-designed tutor could do 1:1, and the Norway recommendation discourages both that usage and the copy-paste “just give me the answer” use the evidence actually warns against.
Really glad you’re reading - and thank you for the comment!
Enjoyed this piece, Andrew! As a fellow parent who uses AI to supplement/elevate my children's education, the studies you laid out are compelling and it's a real shame that the wrong studies are being used to draw the wrong conclusions about what is suitable as an educational tool for children.
A few thoughts:
1. The term "ban" is yet another thing that the media has gotten wrong. Norway hasn't actually banned AI for children (which implies legislation and comes off as more draconian). Rather, the Directorate of Education has provided a national recommendation against the use of AI.
In practice, it may be that many schools would follow the recommendations set out, but to me it's not entirely the same thing as a ban that has been passed into law. And parents who are so inclined (like you and me) can still use AI at home to teach their children.
Whilst one could argue that having such a national recommendation in place can taint the image of AI in general and plausibly lead to fewer parents using it for education even at home, using technology judiciously in education to obtain better results with minimal downsides/side effects (eg. cognitive offloading) has always been a question regardless of the era and not a problem specific to AI, so it looks more to be a user-related matter.
2. "Norway forecloses the one configuration the evidence actually supports: a tutor that withholds the answer and walks a student toward it." - my interpretation of their recommendation against AI for specific ages was that this role of withholding the answer and walking the students towards it is and should still be the teacher's responsibility so it's not necessarily that they've taken that option away entirely.
Looking forward to future posts!
Thank you, Jane - these are sharp, and you’re right on the terminology. A “recommendation” isn’t a “ban,” and I should’ve been more careful with the frame. Why I think it still matters: a national recommendation carries a chilling effect even without the force of law - it shapes what teachers feel licensed to try, which is close to your point about it tainting AI’s image even for home use.
On (2), this is the crux and it's a great pushback. You’re right that withholding the answer and walking a student toward it is the teacher’s job and always has been. My worry is scale - a good teacher does this beautifully one kid at a time, but can’t for 25 at once. That Socratic, answer-withholding role is exactly what a well-designed tutor could do 1:1, and the Norway recommendation discourages both that usage and the copy-paste “just give me the answer” use the evidence actually warns against.
Really glad you’re reading - and thank you for the comment!