view release on metacpan or  search on metacpan

share/SpamAssassin/easy_ham/00851.51d3571c945fd64a210d477de46e14b7  view on Meta::CPAN

References: <DAV68j2OW7MgETDx4e700008892@hotmail.com>
From: Gary Lawrence Murphy <garym@canada.com>
X-Home-Page: http://www.teledyn.com
Organization: TCI Business Innovation through Open Source Computing
Message-Id: <m28z1grenj.fsf@maya.dyndns.org>
Reply-To: Gary Lawrence Murphy <garym@canada.com>
X-Url: http://www.teledyn.com/
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: fork-admin@xent.com
Errors-To: fork-admin@xent.com
X-Beenthere: fork@spamassassin.taint.org
X-Mailman-Version: 2.0.11
Precedence: bulk
List-Help: <mailto:fork-request@xent.com?subject=help>
List-Post: <mailto:fork@spamassassin.taint.org>
List-Subscribe: <http://xent.com/mailman/listinfo/fork>, <mailto:fork-request@xent.com?subject=subscribe>
List-Id: Friends of Rohit Khare <fork.xent.com>
List-Unsubscribe: <http://xent.com/mailman/listinfo/fork>,
    <mailto:fork-request@xent.com?subject=unsubscribe>
List-Archive: <http://xent.com/pipermail/fork/>
Date: 02 Oct 2002 12:25:52 -0400


How is this any different from attaching an Infobot or A.L.I.C.E
through licq's console-hook?  People have been doing that for years,
and for over a decade in IRC and the MUDs.

... and the one thing I think we've learned in all that time is that,
as a help-desk, it doesn't work.  People just don't like talking to
robots, especially when the robots, once confused, become imbeciles.
I think the humans may feel cheated, deceived and tricked when they
discover, as with the Seattle answer, that they are talking to a
machine; there's no real intelligence behind that simulated-friendly
and therefore empty 'thx'.

AIML is clever and cute, but for /practical/ applications as a first
line of technical support?  It's been tried over and over, and while I
/also/ think that it /should/ work, for the most part, people won't
use it.  What's worse, as we make the NL processing more and more
clever, it only means it fails more dramatically; ALICE doesn't just
stumble a bit, it starts to drool.  And ALICE is the best we have.

Like a Dalek: All very impressive when things are going well, but all
it takes to betray the chicken-brain inside is a towel over it's
ill-placed eye, or a spin off the metal surface ;)

In all the prolog-based NL database query systems of the 1980's and
other later chatterbot helpdesk projects like Shallow Red, even
simpler tries like Ask Jeeves, people very quickly know they're
talking to a robot, and the queries anneal to short, truncated and
terse database-like verb-noun or just noun-keyword requests.  

People are just too quick to adapt, and too impatient to forgive a
clunky interface, and for now, especially when the /average/ computer
user still can't type more than maybe 5-10wpm, NL is a painfully slow
clunky interface.

Put it this way: Would you login, wake the bot and ask for the Seattle
weather, or would you do as we /all/ do and just click the weather
icon sitting there on your desktop?

Just for fun, here's an interesting conversation between Shallow Red,
ALICE and Eliza as they decide to play the Turing Game:

http://www.botspot.com/best/12-09-97.htm

-- 
Gary Lawrence Murphy - garym@teledyn.com - TeleDynamics Communications
 - blog: http://www.auracom.com/~teledyn - biz: http://teledyn.com/ -
  "Computers are useless. They can only give you answers." (Picasso)