[ExI] GPT-4 on its inability to solve the symbol grounding problem

Brent Allsop brent.allsop at gmail.com
Sun Apr 9 12:13:56 UTC 2023

Wow, all that is such a great story.  Yes, Gordon, thanks for organizing
that masterpiece, and thanks Gordan and Spike, for relating your experience
of it, and its implications.

And, EVERYONE, thanks for all this effort!!  Do you realize what we are
doing, as a team?  We are specing out a theoretical design of how
consciousness works, and in my opinion, our team is doing a better job than
any team, anywhere in the world.  There is nothing in the peer reviewed
journals that comes anywhere close to what we are doing here.  To the
degree to which it is an accurate design, not only does that tell us what
humans are, it will enable us to predict the near future of humanity, what
engineered uploading to superior brains will be like, and all that.  It
doesn't seem to be true, but there is a LOT of stuff we are all agreeing
on.  We just don't talk about  what we agree one, 99% of our words are just
about the minor details we don't agree with, seemingly endlessly.  AND I
can see some people, including myself, changing the way we think about
consciousness, or at least improve the words we use to describe what we
think, as a result of this conversation.

The goal of the Theories of Consciousness
<https://canonizer.com/topic/88-Theories-of-Consciousness/1-Agreement> topic
is to capture this specification, real time, as it progresses.  The things
we agree on, like it is approachable via science
and conscious knowledge is composed of qualities like redness and greenness
can be specced out in the super camps, and when we find stuff we disagree
on (the nature of these qualities of which consciousness is composed), this
can be pushed down to supporting sub camps, out of the way of the more
important consensus stuff we do agree on.  Then we basically vote on that
stuff (i.e. track consensus, but don't lose track of the minority opinions)
resulting in a ladder of the most popular way to think about things in the
most popular camps, in ever lower levels of detail with decreasing amount
of consensus.  Certainly we all know science (or a new good argument) could
verify one of the minority camps, falsifying more popular competing camps
and forcing the consensus into that camp, this pushing it towards the top
of the consensus ladder as more people jump on board.  We are currently
working on an animation system so you can see the "as of" values animated
over time, showing the progress of the consensus design tree.  Once we have
that animation, showing the progress of the design over time, I think
people will better understand.

On Sat, Apr 8, 2023 at 10:18 PM spike jones via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> *From:* extropy-chat <extropy-chat-bounces at lists.extropy.org> *On Behalf
> Of *Gordon Swobe via extropy-chat
> >…My main takeaway from that intense four month experience was that world
> champion chess players like Kasparov live in a different world. We all
> thought the world team had a good chance to win. We had the benefit of
> computers and expert advice and the supposed power of democracy on our
> side, but in the end, it felt like Kasparov was Mozart on the piano and we
> were children playing on harmonicas.
> -gts
> Gordon I followed that game realtime.  I recall it being a crazy-looking
> thing.  I thought we had real chances for a draw way down until the end,
> before I realized Kasparov was a tempo ahead of us.  (I don’t wish to make
> it sound like I participated at the time: I was too busy at the time so I
> only watched, but I considered myself one of the not-Kasparov part of
> humanity.)  As I recall it felt like Kasparov somehow pulled his magic
> trick in the last ten moves of the game, but I couldn’t find where he
> gained that tempo on us.
> Thanks for organizing that masterpiece.
> spike
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230409/48403092/attachment-0001.htm>

More information about the extropy-chat mailing list