On 11/13/06, Richard Loosemore <[EMAIL PROTECTED]> wrote:
But....
Now you have me really confused, because Searle's attack would have
targetted your approach, my approach and Ben's approach equally: none
of us have moved on from the position he was attacking!
The situation is not that hopeless. ;-)
The key here is not whether an intelligent system uses "symbols", but
how these symbols get their meaning.
Both Searle and GOFAI assume Tarskian (model-theoretic) semantics,
where the meaning of a symbol is its "denotation" in the domain, and
the mapping ("interpretation") is independent of the activities of
system that actually using the symbol, but determined by an outside
observer.
This problem is gone if the meaning of a symbol is defined by, and
used as, the role the symbol plays in the system's experience, which
does not necessarily equal to human sensory experience.
A detailed discussion is in http://nars.wang.googlepages.com/wang.semantics.pdf
Pei
A debate about Searle is not really something I want to get into,
unfortunately, because Searle's claims have changed over the years and
(much, much worse) they were never very coherent even at the beginning.
Broadly speaking Searle would ask you if you are building your AI out
of real neurons, and if you answer "no" he will say it does not
understand anything, it is just an empty shell that lacks something we
have, and that it is not conscious. As far as I can see, that puts all
of us out inthe cold.
When you say that people have moved beyond the "simplistic GOFAI
symbol-as-intelligence idea", that looks like you are talking about
different issues entirely. Perhaps symbol grounding, perhaps other
issues. I think all of us have moved on from most of the "simplistic
GOFAI ideas".
Richard Loosemore
John Scanlon wrote:
> I was referring to the kind of symbol-system hypothesis that Searle's
> Chinese room and Hubert Dreyfus's writings attack, and wondering if
> there were still people following the approach they attacked. So the
> responses I've gotten (and lack thereof) are heartening. Symbols, in
> the form of language, are a very important part of the system (Gnoljinn)
> I'm developing. These symbols will be a crucial component for abstract
> thinking (once I get that far). It's good to hear that researchers now
> have moved beyond the simplistic GOFAI symbol-as-intelligence idea --
> more people with more advanced ideas to share thoughts with.
>
> ----- Original Message ----- From: "Richard Loosemore" <[EMAIL PROTECTED]>
> To: <[email protected]>
> Sent: Sunday, November 12, 2006 4:37 PM
> Subject: Re: [agi] A question on the symbol-system hypothesis
>
>
>> John,
>>
>> The problem is that your phrases below have been used by people I
>> completely disagree with (John Searle) and also by people I completely
>> agree with (Doug Hofstadter) .... in different contexts, they mean
>> totally different things.
>>
>> I am not quite sure how it bears on the quote of mine below, but for
>> the record I don't have anything much against the physical symbol
>> system hypothesis per se (even though I also agree with Stevan
>> Harnad), but I do have a problem with people who think that symbols
>> are passive, structureless objects manipulated by something external.
>> (And my objection is not so much that it is nakedly wrong, as that it
>> diabolically inconsistent with a lot of stuff, and untested).
>>
>> From what you write, I think it was the latter issue that you were
>> referring to.
>>
>> Richard Loosemore.
>>
>>
>>
>>
>> John Scanlon wrote:
>>> I get the impression that a lot of people interested in AI still
>>> believe that the mental manipulation of symbols is equivalent to
>>> thought. As many other people understand now, symbol-manipulation is
>>> not thought. Instead, symbols can be manipulated by thought to solve
>>> various problems that can be solved that way. Intelligence is
>>> primary to the ability to use language -- why we recognize in animals
>>> a certain level of intelligence and sentience. The ability to use
>>> symbols and language at the human level depends on more
>>> sophisticated, specialized functions in the brain than are found in
>>> other earthly species, but the symbol-manipulation is still not
>>> thought -- it is done by thought.
>>> My question is: am I wrong that there are still people out there
>>> that buy the symbol-system hypothesis? including the idea that a
>>> system based on the mechanical manipulation of statements in logic,
>>> without a foundation of primary intelligence to support it, can
>>> produce thought?
>>> This comment from Richard Loosemore made me think about this question:
>>> (I can't tell from this if he supports the symbol-system hypothesis
>>> or not.)
>>> > Whether committed to human-inspired AI, or to anti-human ;-)
>>> > Normative Rational AI, it was always some long-ago
>>> > introspection that was the original source of the ideas that
>>> > are now being formalized and implemented. Even logical,
>>> > rational thought was noticed by the ancient Greek
>>> > philosophers who looked inside themselves and wondered how it
>>> > was that their thoughts could lead to conclusions about the world.
>>> ------------------------------------------------------------------------
>>> This list is sponsored by AGIRI: http://www.agiri.org/email To
>>> unsubscribe or change your options, please go to:
>>> http://v2.listbox.com/member/?list_id=303
>>
>> -----
>> This list is sponsored by AGIRI: http://www.agiri.org/email
>> To unsubscribe or change your options, please go to:
>> http://v2.listbox.com/member/?list_id=303
>>
>
> -----
> This list is sponsored by AGIRI: http://www.agiri.org/email
> To unsubscribe or change your options, please go to:
> http://v2.listbox.com/member/?list_id=303
>
>
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?list_id=303
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?list_id=303
