590
Aww. (lemmy.world)
you are viewing a single comment's thread
view the rest of the comments
[-] dactylotheca@suppo.fi 1 points 6 days ago

I think the Turing test is flawed; eg. ChatGPT playing the part of a doctor is more likely to be empathetic than a real doctor.

The question isn't whether computers can act like humans; the question is can we.

[-] MBM@kbin.run 5 points 6 days ago

My problem with the Turing test is basically Goodhart's law: "When a measure becomes a target, it ceases to be a good measure"

[-] laughterlaughter@lemmy.world 1 points 5 days ago

Your comment is weird. The point of the test is to fool humans. Humans were fooled. What does it have to do with measurement and Goodhart's law?

[-] Mouselemming@sh.itjust.works 3 points 5 days ago

I got a phishing spam pretending to be the USPS the other day. Had the logo and everything, grammar was good, but at the end it said something like "USPS wishes you good luck every day." I know the fucking post office isn't that benevolent! Especially under the present Postmaster General. Block and Report.

[-] laughterlaughter@lemmy.world 1 points 5 days ago

I've heard people say "omg chatgpt so stoopid, not the point of the Turing test" lately.

To me, that's moving the goalpost.

The Turing test is about fooling a human being into thinking he/she is interacting with another human being.

Does this happen? Yes. 100%. Or at least almost 100% (because not everyone is fooled.)

Now, the whole thing about chatgpt being stupid, etc, well.... that's another matter.

Come up with another test. Name it the Goofy test, or whatever.

[-] Isa@feddit.org 2 points 6 days ago* (last edited 6 days ago)

I'd answer that last question with a clear … no!

At least not in general.

this post was submitted on 26 Jun 2024
590 points (97.3% liked)

Crows

668 readers
130 users here now

Bring stories and images of your encounters with crows, ravens, and other corvids. Link articles and anecdotes. Note here your literary references.

founded 1 year ago
MODERATORS