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

[Giovanni Santostasi]

I wanted to study this project  and now after this discussion, I'm finally
doing it:

https://www.wolframphysics.org/

I read the first few pages of the book and I finally found what I mentioned
in my previous messages: a bootstrapping approach to building a language.
Not only Wolfram is trying to build a language from very simple rules
(often adding plenty of self-referential rules) but an entire freaking
universe. It is able to create space and time. He is able to derive the
invariance of relativity.

So the issue if you can build a language from very simple rules and a few
initial abstract objects like integers (and if you have an integer you have
all of them) it is not an issue any longer given it seems you can build an
entire universe from this protocol.

Gordon, you should explore this website and maybe get the book so you can
have a feeling of how this is done. It is all about relations and
interactions between a few initial objects and rules on how to make updates
on the relationships. This is exactly what the NLM do with their NNs and in
fact what our brains do too.

Every single experience, memory, idea, or word is a sort of graph or
connected pattern in our brain. All that the brain cares about is the
sequence of activation: neuron 2, followed by neuron 1, followed by neuron
5. That is a chair or whatever. Or at least some aspects of the chair,
other connections, short and long across the brain create other
associations like a chair being made of wood or being something you sit on.

Meaning is built in relating this activation pattern to another activation
pattern, for example knowing that a chair is smaller than a house and it
can be inside a house or that a person (another activation pattern) can sit
on the chair or the chair is made of wood (another activation pattern).

To build meaning you don't need to know what wood is but simply that a
chair is made of wood and wood is the material that threes are made of and
threes are plants that are one of the forms of living beings and so and so
on.

At no point, you need to refer to any real object in the real world, all
that you care about is the relations between these objects that can be
identified by specific and unique activations patterns.  You can do this
with particles and forces of nature and you can do this with a natural
language like English. This is exactly what the NLMs have done. It is just
a bunch of weights in a NN and activations patterns in these NNs, exactly
like in our brains or in the universe.

I don't understand how an intelligent person doesn't get this. I'm serious.

Giovanni










On Wed, Apr 5, 2023 at 8:49 PM Jason Resch via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

>
>
> On Wed, Apr 5, 2023, 11:26 PM Gordon Swobe <gordon.swobe at gmail.com> wrote:
>
>> Frankly I am dumbfounded and flabbergasted that any intelligent person
>> would question my statement  "Words mean things. In the absence of those
>> things that they mean, they have no meanings."
>>
>
>
> "Words mean things" -- no disagreement here
>
> "In the absence of the things they mean, they have no meaning" -- This I
> disagree with. If two English speakers survived while the rest of the
> universe disappeared completely, the two speakers could still carry on a
> meaningful conversation. Their words would still mean things to them. As
> long as there's a brain with an appropriate wiring to process words and
> comprehend the network of relations each word has with other words, there
> will be meaning. Meaning exists within the mind of the speaker, the
> presence or absence of an external universe is irrelevant from the point of
> view of the mind (which for all it knows could be dreaming, deluded, or in
> a vat or sim).
>
>
> Jason
>
>
>> How do you all think you communicate here on ExI or IRL? You use words
>> that mean things to you and which you expect will mean very similar things
>> to others. The word-symbols that you write or utter are merely the vehicles
>> for the meanings. Words without meanings are no more than, well,
>> meaningless nonsense.
>>
>> -gts
>>
>>
>>> _______________________________________________
> 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/20230405/6f7f874f/attachment.htm>