I might be a bit behind the trend, but tried chatgpt today.
me: I have questions about #guile programming. Is there an equivalent to OCaml's `for_all` function?
AI: In Guile, you can use the `every` procedure to achieve the same behavior as OCaml's `for_all` function.
I asked something similar (but less clear, I'll admit) on IRC two days ago, and I was first suggested how to reimplement it before someone talked about "every" (:
@roptat
Tried it with common lisp to write a simple redis client. It kept throwing thin wrappers around libs that didn't exist, even though I asked it to use quicklisp.
@roptat
For me the experience was of hiring a junior dev who does nothing more than copy-paste random stackoverflow chunks.
Wonder if it's the same for other domains of knowjedge where some level of expertise is required.
@seeg @roptat (sorry for replying into a very old thread) my experience with asking chatgpt some simple astrophysics questions was the same: it can provide common responses to basic questions (i.e. why the sky is dark at night?), but they are often quite inaccurate. When I asked it to explain Olbers paradox (which is, basically, why the sky is dark if surface brightness does not depend on the distance) to me, it did so eagerly, however, it was unable to make a connection between its own responses and being asked "so, why the sky is dark, then?” replied with the same inaccurate answer as it did at step 1.
High school kids whom I explained this topic were typically able to see the point, so, basically, human intelligence still wins over some overtrained ai :)
