[ExI] all we are is just llms was: RE: e: GPT-4 on its inability to solve the symbol grounding problem

[spike at rainier66.com]

From: Giovanni Santostasi <gsantostasi at gmail.com> 
…
Subject: Re: [ExI] all we are is just llms was: RE: e: GPT-4 on its inability to solve the symbol grounding problem

 

>…Also, it seems Bender and Mitchel are quoting and re-tweeting each other often. I guess AGI haters unite! Nice coincidence? Gordon, what is the name of this club?...

 

 

Hmmm, those who disagree with the notion that ChatGPT has become conscious are not necessarily AGI haters.  I use ChatGPT nearly every day for something and agree it is a marvelous tool.  I will agree that I had never thought of consciousness as a continuum, because we humans don’t really have that.  We sleep, we wake, and we are conscious.  I recognize that there can be something like 0.1 human AGI which can be a little conscious and do some things way better than humans.  This is not AGI hating.

I will also note that playing the Turing Test game has become quite popular.  If Turing had lived long enough to see this day he would have said “Well chaps, we are there.  This software is human-level machine intelligence.”  By Turing’s own criterion we are there.  If Turing had lived, he would be 111 years old, and he would STILL be smarter than I am now.

However… I have come to realize that we can have human level general intelligence without actual human level consciousness.  This is not AGI hating at all, but rather a realization that consciousness and intelligence are concepts way more difficult to define than I had previously thought.  I once believed I fully understood what those terms mean, but now I realize they are far more subtle than I had imagined.

spike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230420/bab78b75/attachment.htm>