Hi Brian. Thanks for posting the piece on the [AI in Ed] email group and leading me to your Substack.
Your distinction between AI as a "brain booster" versus a "brain drain" perfectly aligns with a core theme I’ve brought up in my writing on Medium.
This brings up a key question for the education space: What if we could use prompts to actually force the student to engage in the very kind of "System 2" thinking you advocate for? Have you considered how techniques exist that make the AI show its work, compelling it to break down problems step-by-step instead of just giving a final answer?
If we teach metacognitive prompts, frame "trial and error" AI process, which are fundamentally similar to how humans learn. It seems the challenge is less about AI being a shortcut and more about guiding it to become a catalyst for deeper learning.
Agreed! I think finding ways to use AI to actually engage System 2 thinking is the most promising for current-gen AI. Mike Caulfield (here on Substack) has done a lot with that, showing how to craft prompts in ways that are infinitely more useful (and less prone to falling for errors) than default prompting. I think there's a lot of research to be done into what kinds of interactions and UI/UX design lead to genuine skill and knowledge growth (as opposed to, say, in-the-moment efficiency but long-term atrophy).
Mike’s name sounds familiar. I’ll definitely check it out. I agree there is a lot more for this space. I’ve been trying to outline some thoughts for a post.
Hi Brian. Thanks for posting the piece on the [AI in Ed] email group and leading me to your Substack.
Your distinction between AI as a "brain booster" versus a "brain drain" perfectly aligns with a core theme I’ve brought up in my writing on Medium.
This brings up a key question for the education space: What if we could use prompts to actually force the student to engage in the very kind of "System 2" thinking you advocate for? Have you considered how techniques exist that make the AI show its work, compelling it to break down problems step-by-step instead of just giving a final answer?
If we teach metacognitive prompts, frame "trial and error" AI process, which are fundamentally similar to how humans learn. It seems the challenge is less about AI being a shortcut and more about guiding it to become a catalyst for deeper learning.
I rambled a bit here, but hope we can connect.
Agreed! I think finding ways to use AI to actually engage System 2 thinking is the most promising for current-gen AI. Mike Caulfield (here on Substack) has done a lot with that, showing how to craft prompts in ways that are infinitely more useful (and less prone to falling for errors) than default prompting. I think there's a lot of research to be done into what kinds of interactions and UI/UX design lead to genuine skill and knowledge growth (as opposed to, say, in-the-moment efficiency but long-term atrophy).
Mike’s name sounds familiar. I’ll definitely check it out. I agree there is a lot more for this space. I’ve been trying to outline some thoughts for a post.