GPTChat prompt: Write some Dyalog APL code to perform a left-associative division over an integer vector.

After informing that ⍴ isn't assignable, it stripped out that and the following line.

However, after then informing it that this doesn't perform any scan operation at all, it bailed: "An error occurred. If this issue persists please contact us through our help center at help.openai.com."

It's pretty terrible act producing actual APL code, but asking about algorithmic choices in APL, prodding along when you know it's wrong, does end up getting it to say things that sound quite plausible.

Hah. This explains a lot: jsoftware.com/papers/eem/storyofo.htm

History of Dyadic ○ and why it's "arbitrary and slightly distasteful".