Discussion about this post

User's avatar
Yuvraj Sarda's avatar

> It’s funny: if you were to ask one of our students, “Hey, tell me about your AI tutor!” they’d probably look at you with a blank face.

Well, maybe this is a sign that what you’re calling is not an AI tutor? At least not in the conventional sense. What you’re describing sounds just like spaced repetition or adaptive algorithms. Yes, plausibly you could argue that they fall in the broader umbrella of AI (eg: all algorithms are AI) but the choice of term doesn’t really help anyone understand what the thing you’re selling is.

From the title, I was hoping you’d share more about the mechanics of the “AI” tools you use at Alpha. What algorithms? What is the approach? Any LLMs? What is the UX of using the tools exactly like? But the essay didn’t really go into it.

Heith A. Wetzler's avatar

This is a heck of a leap to take from a two week study of less than 200 students in a highly conditioned environment.

I have no doubt that the EdTech industry will capitalize on this and make billions. Despite the decades of over-promising on the benefits of edtech, THIS technology will be the one to do the trick. But as with all attempts at shortcuts and half measures, this optimism neglects a whole lot. For example, what about the negative effects of gamification on young people’s attention spans and dopamine production. It’s just a matter of time before these are weaponized just as social media and smartphones have become. What about the increasing anxiety, mental heath issues associated with life spent more in the digital vs the actual world? Or how will this account for the real problem of thousands of teens who replace AI for real human relationships?

You say that we must adapt, the earlier the better. But the early adapters of edtech like iPads in classrooms have done nothing but produce more and more abysmal results and greater dependence upon technology for basic things like ideation and word selection.

I’d love to see the links to the other studies you cited. I’d love to see a comparison of this study with other previous studies. Or maybe consider that the researchers themselves in their conclusion do not claim this will be a universal result in every context. Do we know which contexts won’t work as well with this format?

Too many questions and promises, too few answers at this point for me to get on board.

4 more comments...

No posts

Ready for more?