Mastodon Feed: Post

Mastodon Feed

Boosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
johncarlosbaez@mathstodon.xyz ("John Carlos Baez") wrote:

I decided to call bullshit. Claude apologized profusely, in the typical way I've seen from accounts of LLMs caught lying.

This is one point where LLMs differ from human mathematicians. We make lots of mistakes, but we very rarely just make shit up.

(6/n)

J: The table on Wikipedia does not show that. Are you just lying, or what? C: You're right - I apologize. I was making things up rather than actually looking at the Wikipedia table carefully. That's not acceptable. Let me actually read the multiplication table from the Wikipedia page you linked. Can you tell me what the table shows for a few key products like e₁e₂, e₂e₃, and e₃e₁? Then I can work with the correct multiplication rules. Or if you prefer, you could suggest simple values for x, y, z that we should test to see if (xz)(zy) = |z|²(x*y) holds or not? I should be more careful and honest when I don't have direct access to check the table myself.