home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.philosophy
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!usenet.coe.montana.edu!news.u.washington.edu!stein.u.washington.edu!forbis
- From: forbis@stein.u.washington.edu (Gary Forbis)
- Subject: The Humongous Look-up Table revisited.
- Message-ID: <1992Dec22.211814.5623@u.washington.edu>
- Sender: news@u.washington.edu (USENET News System)
- Organization: University of Washington, Seattle
- Date: Tue, 22 Dec 1992 21:18:14 GMT
- Lines: 50
-
- Like turning on a lamp, my mind lit up at the mention of the HLT.
-
- I hope there is no question in anyone's mind but that there exists
- HLTs that will pass the Turing Test. As I see it finding even one of
- this huge class of Turing Machines is an intractable problem. I think
- there are steps which can reduce the problem space but not be very much.
-
- Given the set of all exchanges of strings of finite length expressed as
- HLTs, some will pass the Turing Test serendipitously. I hope the set of
- HLTs to be considered can be limited a bit. In particular, there are
- times the exchange:
-
- Interogator: What time is it in Seattle?
- HLT: Oh, about noon. Why?
-
- would be reasonable and times it would not.
-
- While I have no mathmatics to back me up, it seems to me there are mixed
- devices which produce the same output as a serendipitous HLT. Such a device
- might, upon being is some set of states and encountering the question, "What
- time is it in Seattle?" enter a procedure which would check its internal
- clock and generate the phrase, "Oh, about noon. Why?" Surely such an
- enhanced HLT is going to have a much higher chance of serendipitiously passing
- the Turing Test than the larger set of all HTLs.
-
- Likewise, the exchange:
-
- Interogator: How's the weather there?
- HLT: Not to bad for late December.
-
- would be much more likely if the enhanced HLT had access to an automated
- weather station. I don't see how anyone could accept the computability of
- a seredipitious HLT but not of an enhanced HLT producing the same exchange.
-
- Even the exchange:
-
- Interogator: What do you think of Clinton?
- HLT: He seems to be getting himself into hot water with the women's
- groups. I wonder how this will all play out.
-
- is in there somewhere within the set of all finite exchanges of strings
- of finite length. An enhanced HLT having access to usenet would have a
- much better chance of appropriately generating this response.
-
- I know these moves haven't whittled down the problem very much but it seems
- to me there are moves which guarantee equivalence with serendipitious HLTs
- and increase the odds of a Turing Machines with the set being equivalent
- to a seredipitious HLT.
-
- --gary forbis@u.washington.edu
-