[ExI] Bender's Octopus (re: LLMs like ChatGPT)

[Jason Resch]

On Sat, Mar 25, 2023 at 12:03 PM William Flynn Wallace via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

>
>
> On Fri, Mar 24, 2023 at 2:16 PM Jason Resch via extropy-chat <
> extropy-chat at lists.extropy.org> wrote:
>
>>
>>
>> On Fri, Mar 24, 2023, 3:00 PM William Flynn Wallace via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>> Would anyone dare to give a definition of 'understanding'?
>>> Please cite what epistemologies you are using.  bill w
>>>
>>
>> The difficulty is that "understanding" is adjacent to knowledge and
>> knowledge is adjacent to consciousness. All these are quite difficult to
>> define but I will attempt by best:
>>
>> "Understanding" is knowledge concerning the relations or workings of
>> something.  (just why do we just 'standing under' something to represent
>> knowledge?)
>>
>
Good question, I was not aware of the etymology of this (nor did I even
think to question the obvious connection to those two words), but I found
this:
https://en.wiktionary.org/wiki/understand#Etymology
It seems the basic idea of "stand under" is to suggest subordinacy to
another thing. In this case, the understanding derives from the higher
principle that is understood, so it the understanding "stands below" that
thing which is understood.


>
>> "Consciousness" is possession of knowledge. (since no creature has a
>> blank mind, then all are conscious?)
>>
>
I think so, yes.


>
>> "Knowledge" is a true belief.  (true according to what epistemology?
>> empiricism?  authorities?  intuition? reason?)
>>
>
This definition of knowledge traces its roots to Plato's Theaetetus.


>
>> "Belief" I have great difficulty defining, but I would say it is a mind
>> state correlated with some proposition.  (I would say that it is
>> something we think of as knowledge but not based on empiricism but rather
>> on faith)
>>
>
In many cases, but not all. For example: the belief that the sun will rise
tomorrow could be said to be based on empiricism, but it is still a belief.


>
>> "Truth" is undefinable, as proved by Tarski.  (again, true is something
>> we accept according to our personal epistemology)
>>
>
Without getting into whether there are objective truths or not, whatever
truth is, even within rigorously defined mathematical frameworks, is not
something that can be given by a definition. From Gödel, we know that true
statements are not just those that can be proven, so "provable" is not
sufficient to define truth. How then are we to define true statements?
Things only get messier if we attempt to extend definitions of truth beyond
mathematics.


>
>>
>> I don't regard these as corrections, but just my ideas.   bill w
>>
>>
I appreciate them, thank you. :-)

Jason


>
>>
>>>
>>> On Fri, Mar 24, 2023 at 1:40 PM Jason Resch via extropy-chat <
>>> extropy-chat at lists.extropy.org> wrote:
>>>
>>>>
>>>>
>>>> On Fri, Mar 24, 2023 at 1:21 PM Gordon Swobe via extropy-chat <
>>>> extropy-chat at lists.extropy.org> wrote:
>>>>
>>>>>
>>>>>
>>>>> On Fri, Mar 24, 2023 at 2:12 AM Stuart LaForge via extropy-chat <
>>>>> extropy-chat at lists.extropy.org> wrote:
>>>>>
>>>>>
>>>>>> But really the meaning of words are quite arbitrary and determined
>>>>>> by
>>>>>> the people who use them. Thus the referential meanings of words
>>>>>> evolve
>>>>>> and change over time and come to refer to different things
>>>>>
>>>>>
>>>>> I agree this is a reason for many human miscommunications, but the
>>>>> speaker understands his words to meaning *something* and the hearer
>>>>> understands those words to mean *something*.
>>>>>
>>>>> As a computational linguist, Bender is on our side.  She is obviously
>>>>> very excited about the progress these language models represent, but is
>>>>> reminding that the models do not actually understand words to mean anything
>>>>> whatsoever.
>>>>>
>>>>>
>>>>
>>>> What's her evidence of that?
>>>>
>>>> Jason
>>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230325/082fadf4/attachment.htm>