<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);" class="elementToProof">
These are interesting ideas. As long as a question can be answered by referencing written text, some LLM is likely to succeed. Basically, all the required information is embedded in its natural domain. If we want machines to "understand" the world in a non-textual
way, we need to come up with questions that can only be answered by referencing knowledge outside the textual domain. This is of course a bit unfair if the AI that has no opportunity to be embodied and thus learn that way. It also seems to be very difficult
of us to imagine a question presented symbolically that can't be answered entirely with symbolic knowledge.</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);" class="elementToProof">
<br>
</div>
<div>The creation of quasi-novel artwork by stable-diffusion networks might be an example of operating outside the symbolic domain. This is a bit vague, since image representation in machines is ultimately a finely-grained symbolic representation (RGB values
over discrete positions).<br>
</div>
<div id="appendonsend"></div>
<div style="font-family:Calibri,Arial,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
<br>
</div>
<hr tabindex="-1" style="display:inline-block; width:98%">
<div id="divRplyFwdMsg" dir="ltr"><font style="font-size:11pt" face="Calibri, sans-serif" color="#000000"><b>From:</b> Connectionists <connectionists-bounces@mailman.srv.cs.cmu.edu> on behalf of Prof Leslie Smith <l.s.smith@cs.stir.ac.uk><br>
<b>Sent:</b> Friday, March 10, 2023 3:45 PM<br>
<b>To:</b> Adam Krawitz <akrawitz@uvic.ca><br>
<b>Cc:</b> Connectionists List <connectionists@cs.cmu.edu><br>
<b>Subject:</b> [EXTERNAL] Re: Connectionists: Chomsky's apple</font>
<div> </div>
</div>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt">
<div class="PlainText">Dear all:<br>
<br>
I'm beginning to think that we are looking at this from the wrong end: the<br>
issue isn't about what/whether ChatGPT understands, but about what we mean<br>
by "understanding" in a human or an animal, and what we might mean by<br>
understanding in a machine.<br>
<br>
If I say: I'll drive to London, it's clear that (a) I have access to a car<br>
(b) I can drive etc. But I may or may not understand how the car works. I<br>
may or may not understand the nature of the frictional forces that allow<br>
the wheels to move the car. I may or may not understand the chemistry that<br>
allows she internal combustion engine/battery to operate. I (and<br>
presumably the person I am talking to) has a model of understanding of<br>
driving cars that suffices for our needs.<br>
<br>
In other words, out "understanding" relates to the activities we want to<br>
do, activities that are about our operation/interaction in our<br>
environment. So we often use simple models that suffice for our<br>
activities and interactions. Our understanding is systematic, but may well<br>
be wrong, or (more likely) just sufficient for our purposes (I know I need<br>
to put petrol/gas in the car, or I need to charge the battery) rather than<br>
complete (*).<br>
<br>
Our understanding clearly works entirely differently from ChatGPT, (and I<br>
agree with Richard Loosemore that ascribing a human sort of understanding<br>
to ChatGPT is not appropriate).<br>
But if we want to use the same terms to describe machines and humans, we<br>
should really start by deciding what these terms mean when applied to<br>
humans.<br>
<br>
(*) In fact out models are never complete: they rely on concepts like<br>
solidity, fluidity, electrical current, gravity, light, etc., concepts<br>
that we understand sufficiently for everyday usage. Completeness would<br>
imply a full physics that went down to subatomic/quantum levels!<br>
<br>
Adam Krawitz wrote:<br>
><br>
>> ChatGPT's errors reveal that its "understanding" of the world is not<br>
>> systematic but rather consists of patches of competence separated<br>
>> by regions of incompetence and incoherence.<br>
><br>
<br>
</div>
</span></font></div>
</body>
</html>