From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!sarah!cook!psinntp!psinntp!scylla!daryl Mon Mar  9 18:35:59 EST 1992
Article 4333 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!sarah!cook!psinntp!psinntp!scylla!daryl
>From: daryl@oracorp.com (Daryl McCullough)
Subject: Re: Monkey Room
Message-ID: <1992Mar6.163402.4993@oracorp.com>
Organization: ORA Corporation, Ithaca, New York
References: <9203031955.AA11770@ucbvax.Berkeley.EDU>
Date: Fri, 6 Mar 1992 16:34:02 GMT

GUNTHER@WMAVM7.VNET.IBM.COM ("Mike Gunther") writes:

> Suppose we are given a sealed room + teletype setup which has passed a
> Turing Test.  We open the room and find only a monkey, hitting
> teletype keys at random.  It just so happens that the monkey's
> keystrokes produced "intelligent" conversation up until the time the
> room was opened.

> This thought-experiment seems to contradict several ideas-- the Turing
> Test, behaviorism, functionalism, and the Systems Reply for starters.
> Any comments?

It doesn't contradict any of these ideas. In the case of the Turing
Test, behavior is simply taken as *evidence* for intelligence. If the
system is nondeterministic (as you have assumed in the case of the
room full of monkeys), no amont of evidence can be consideredd
conclusive, because you never know whether the behavior you have
witnessed is typical, or a fluke. However, it still works as a
probabilistic indicator of intelligence, since the odds are pretty low
that monkeys would produce an intelligent conversation.

As for behaviorism and functionalism, there, the criterion is not that
any *single* behavior seems intelligent, but that every possible
behavior must be intelligent (you could weaken that, in the case of
nondeterministic systems, to "with very high probability, the behavior
will be intelligent"). The monkeys may occasionally produce
intelligent behavior, but it is not typical, and so the room full of
monkeys is not functionally or behaviorally equivalent to an intelligent
being.

Daryl McCullough
ORA Corp.
Ithaca, NY









