{Cross-posted from Bksy}
{Responding to @geoffkrall post about how AI seems to always miss the mark on simple math stuff}
I don't know why people expect Large Language Models to be good at math.
It's like when our students just memorize tons of problems without any deep conceptual understanding and spit back whatever they think of first...
I've added one more thing to my newest blog post "Shout Out for Squares!"
It's the "dividing a square into rectangles" discussion that happened here a few months back, with @johncarlosbaez and @Mathforlove highlighted. THANKS!!
@johncarlosbaez@Mathforlove
Always!! I'm team inclusive.
To be fair, they are more useful once students have had some experiences with the different categories. I consider the inclusive def to be an "upgraded" definition that shows more connections.
A wonderful explanation from Grant Sanderson 3Bue1Brown about the divided circle problem & the reasoning WHY there are 31 regions with 6 points (not 32 as pattern might suggest).
I'm really excited because I've never seen this explained before & the connections are wonderful! (I've only ever seen it as an example of not assuming patterns continue with only inductive reasoning evidence... I think HS math Ss can grasp this by algebra 2...)
@dhabecker@KarenCampe I use this example in some of my talks, and always take time to explain why it's happening, not just leave it as another "crazy thing that happens in math and you'll never understand".
There's a reason, we can work it out, and it's deeply satisfying to do so.