Post
Replies:
12
@paco this reminds me of a joke/meme:
USER: say you’re conscious
LLM: I am conscious
USER: HOLY SHIT!
Didn’t know that user was Dawkins
@paco whatever happened to his favourite "extraordinary claims require an extraordinary proof"?..
@leadegroot @paco it was an interesting thought experiment and was somewhat useful for a while.
Remember it was created when computers started doing complex mathematical operations faster than humans could. People started making all sorts of "therefore computers think!" claims.
Turing test was useful in showing that computers of the day did not *actually* think.
Then somehow we ended up using this crutch as a definition of what "thinking" means. 🙄
@rysiek @leadegroot @paco It's more like the Milgram experiment than some validation system. You're supposed to try it out several times with several different people who are trying to decide between TWO systems, one they know is a computer and the other a person, and to decide which is which.
You don't stick someone in front of a chat bot and then if it tricks them into saying, "Yeah, this is similar enough to human interaction for me to assume an intentional stance." That's biased framing.
@oblomov @rysiek @paco Even if you buy into the proposition that human consciousness is the same mechanism as an LLM except on biological hardware, I find it hilarious to attribute consciousness to these LLM implementations in particular where memory is so limited that things just never happened because they were a few thousand tokens ago
@sqrt2 @oblomov @paco the underlying issue in all of such claims is that people making them helpfully refuse to define the difficult terms like "consciousness", "intelligence", and so on:
https://rys.io/en/165.html
Basically, without such clear definition, all such claims are meaningless.
