Humanist Discussion Group, Vol. 39, No. 100.
        Department of Digital Humanities, University of Cologne
                      Hosted by DH-Cologne
                       www.dhhumanist.org
                Submit to: humanist@dhhumanist.org




        Date: 2025-07-25 19:13:55+00:00
        From: James Rovira <jamesrovira@gmail.com>
        Subject: Re: [Humanist] 39.97: repetition vs intelligence: on LLMs

The majority of Gabriel Egan's responses to Tim Smithers are irrelevant to
the question of the possibility of either machine consciousness or the
ability of a machine to comprehend meaning. None of these responses ever
get to the point. He can disagree with the details of Tim's claims without
invalidating his claims, and if he were to invalidate Tim's claims
completely, that still would not support the opposite thesis. Disproving
one thesis is not proof that the opposite is true.

However, Gabriel does get directly to the point here with the following
predictable rebuttal. I am glad he brought it up:

"Looking at what an LLM does with manipulating weights in a neural network
is like looking at what neurons do in the brain: it is looking for the
comprehension of meaning at the wrong level. The thing we call
representation is emergent in the behaviour of the system as a whole, and
it matters not what the bits inside the system are doing to create that
emergent behaviour."

I agree with the idea that looking at neural activity in a human brain
won't disclose the existence or nature of consciousness any more than
looking at the electrical activity in a computer's circuit board would tell
us about the functioning of a computer on any level. Or, if you prefer this
phrase to consciousness, fine: "comprehension of meaning." We can't look at
that level to deduce comprehension of meaning. I agree.

And I agree that we need to look at the nature of "the system as a whole"
to understand it and the possibility of either consciousness or the
"comprehension of meaning."

And I have already responded to that point.

The nature of the "system as a whole" in human beings and in literally any
other organic being is fundamentally different from that of computers. It
is qualitatively different in very fundamental ways, so that the two
systems as a whole cannot be compared at any meaningful point beyond being
electrical on some level. It is currently a divide that cannot be bridged,
not until computers are fully organic or are built in ways that simulate
organic beings. I'm not talking about body shape and mobility. I'm talking
about non-stop actual sensory input that defines the mental space and about
having actual skin. The ability to feel pain, not just take measurements of
heat and cold and light and sound. I've already mentioned human sense
perception along these lines and how it is fundamentally different, the
mind being distributed throughout the body, and its constant interaction
with a physical environment, etc.

We know that the comprehension of meaning exists in organic beings given
these preconditions that do not exist in computers.

Therefore, the safe assumption is that computers do not comprehend meaning.
If you want to assert the fact of the comprehension of meaning or even
the *possibility *of the comprehension of meaning in a computer system, you 
have to provide your own independent evidence of that, and it can't be just 
behavioral. Human beings project human qualities onto inanimate objects *all 
of the time*. That doesn't tell us anything about the inanimate object, however.

Jim R


_______________________________________________
Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php