[ExI] ChatGPT-4 gets the joke, almost

[Gordon Swobe]

I’m not hell-bent on anything, Giovanni, except getting the truth out that
these language models are no different in principle from any other
application, most similar to an autocomplete function in a word-processing
app, just vastly more sophisticated and deeply trained on massive amounts
of text such they can do remarkable things including the mimicry of someone
writing in the first person. They do this so well that gullible people are
fooled into believing they are interacting with conscious beings.

Riddle me this, genius Giovanni,

Why do you suppose Sam Altman and Ilya Suskever, founders of OpenAI,
believe the only good test of a conscious LLM would involve training it on
material completely devoid of references to first person experience?

To put this another way, why you do believe the founders of OpenAI think we
cannot say GPT-4 is sentient while you preach otherwise and harrass and
annoy me for agreeing with them?

-gts
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230425/001b16e3/attachment.htm>