Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!tank.news.pipex.net!pipex!peer-news.britain.eu.net!newsfeed.ed.ac.uk!dcs.ed.ac.uk!flugga.dcs.ed.ac.uk!iic
From: iic@dcs.ed.ac.uk (Ian Clarke)
Subject: Re: Should we be polite to computers?
X-Nntp-Posting-Host: flugga.dcs.ed.ac.uk
Message-ID: <DnJrq4.Fv2.0.staffin.dcs.ed.ac.uk@dcs.ed.ac.uk>
Sender: iic@flugga.dcs.ed.ac.uk (Ian Clarke)
Organization: Department of Computer Science, University of Edinburgh
X-Newsreader: xrn 8.00
References: <DMxnDz.FBC.0.staffin.dcs.ed.ac.uk@dcs.ed.ac.uk> <4g9hbc$bqv@signal.dra.hmg.gb> <825289543snx@sv.span.com>
Date: Thu, 29 Feb 1996 17:10:52 GMT
Lines: 54

In article <825289543snx@sv.span.com>, Shez <Shez@sv.span.com> writes:
> bam@hermes.mod.uk (Brian A. Mellor) writes:
> 
> > In Wizard of Oz experiments, users have bene found to be more polite to
> > simulated computers than to real human operators.
> 
> By polite do you mean using noise words like "please" and "thankyou"? If
> so this result is surprising, as research by British Telecom in the
> 1980's produced the opposite result. Where people thought they were
> talking to a machine they spoke more tersely with fewer noise words.
> This eased the task of producing a speech recognition system as people
> were voluntarily constraining their vocabulary when they thought they
> were talking to a machine.
> 
> The vocabulary used by the machine will of course have an effect on the
> results, as would how "machine-like" the voice was. IIRC, BT's tests
> compared identical dialogue scripts and used humans throughout: for the
> "machine" trials they simply distorted the voice through a synthesiser
> to fool the subjects into thinking they were talking to a machine.
> 
> However, the probable reason for the limited dialogue is the subjects'
> low expectation of the computer's linguistic ability. As people's
> expectations of computers' abilities increases, it's likely they will
> use increasingly complex dialogue. It's possible that the work BT did in
> the mid 1980's is already out of date in this respect, given the
> prevalence of talking computers in SF films & TV shows these days. Also
> I note that unlike BT's experiment, actual voice services use sampled
> speech which is of human-quality except for the lack of appropriate
> intonation. At the time BT did its research I don't think there were any
> voice services (apart from the speaking clock), and the distorted speech
> used in their experiments probably sounded pretty Dalek-like.
> 
> -Shez.
> 

When I posted the article which started this thread, I was really
considering the benifits politeness may have to the humans rather
than the computers, ie. the psycological advantages of treating
computers in a more human way (assuming of course that they can
understand such interaction).  I feel that people may be more likely
to treat other people as they do computers in a world where computers
are used even more than at present.  It may be a simplistic viewpoint
but I think it is applicable on several different levels.  I have
noticed that in general people who use computers more tend to converse
with other people in a more 'computer'-like manner, in that they rarely
(if you will forgive the analogy) 'beat around the bush' in conversation.
If speech could be used with computers I think that this effect would
become many times more prominent, which could be dangerous for society
as we know it.

-- 
|IAN CLARKE        I.Clarke@sms.ed.ac.uk  "..until human voices wake  |
|                  iic@dcs.ed.ac.uk       us, and we drown" - Shelley |
|                  I.Clarke@ed.ac.uk      ianc@aisb.ed.ac.uk          |
