Valentina Poletti <[EMAIL PROTECTED]> wrote:
> Suppose now the book that the guy in the room holds is a chinese-teaching
> book for english speakers. The guy can read it for as long as he wishes, and
> can consult it in order to give the answers to the chinese speakers
> interacting with him.

Great point! Suppose that the rules of the room include the ability to
translate the conversations into English and the rules occasionally do
this.  The man inside the room could eventually learn to understand
the Chinese language.  Does this prove that if the rules include
Chinese to English translations that rule-based systems and computers
can think?

Only by examining a problem from different vantage points, and
questioning how these new insights are relevant to a problem, can
greater understanding of the problem can emerge.  The over reliance on
an initial reaction to a problem or a stock solution method may be
indicative of a tendency to see everything only in the terms of the
initial presumptions that one brings to the study.

The question of whether a computer program, which only follows
instructions, can ever exhibit free-will, has to be seen in the
unpredictability, potentiality and sheer range of possibilities that
interactive learning can produce.  An accurate prediction of how the
program will react at a later point cannot always (or even usually) be
determined because the program can learn something new in that time.
And because of the immense number of possible reactions that are
feasible for a computer program, it is doubtful that anyone could
actually make an accurate and precise prediction of what an
intelligent program would do in the next few minutes even if it didn't
learn something new.  So even though the program may not show free
will in the simplistic sense according to some presumption of what
free will is, it can show an immense range of actions that may be
quite insightful about the IO data environment which is operates in.

An AGI program is both deterministic and non-deterministic relative to
precision and future range of a prediction about what the program will
do.

And the Chinese Room problem, while interesting, cannot be interpreted
by a simple presumption about it.

Jim Bromer



On Fri, Aug 8, 2008 at 6:18 AM, Valentina Poletti <[EMAIL PROTECTED]> wrote:
> Let me ask about a special case of this argument.
>
> Suppose now the book that the guy in the room holds is a chinese-teaching
> book for english speakers. The guy can read it for as long as he wishes, and
> can consult it in order to give the answers to the chinese speakers
> interacting with him.
>
> In this situation, although the setting has not changed much physically
> speaking, the guy can be said to use his free will rather than a controlled
> approach to answer questions. But is that true? The amount of information
> exchanged is the same. The energy used is the same. The person believes his
> decision are now guided by free will, but truly they are still guided by the
> book: if the book gives him the wrong meaning of a word, he will make a
> mistake when answering a chinese speaker. So his free will is just an
> illusion.
>
> The main difference in this second context is that the contents of the book
> were transferred to the brain of the person, and these contents were
> compressed (rather than consulting each case for what to do, he has been
> taught general rules on what to do). The person has acquired understanding
> of chinese from the book? No, he has acquired information from the book.
> Information alone is not enough for understanding to exist. There must be
> energy processing it.
>
> By this definition a machine can understand.
>
>
> ________________________________
> agi | Archives | Modify Your Subscription


-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=8660244&id_secret=108809214-a0d121
Powered by Listbox: http://www.listbox.com

Reply via email to