[alicebot-developer] ALICE mention in this week's edition of New Scientist

[Helio Perroni Filho]

--- "Dr. Rich Wallace" <drwallace@alicebot.org> wrote:

>
http://www.newscientist.com/channel/info-tech/mg18624961.700
> is the url.

Well, Mr. Mullins story is amusing to read, and it did
make me much more likely to check the current state of
Cyc and give it a try (after googling for Cyc's
homepage, since the story itself provides no link).
However, I can't help but feel this story is biased.

* First off, the story gives a report of an experience
with a chatterbot, but what about an example of Cyc
performance? And by the way, on what base is
Intellibuddy "one of the leading artificial
intelligences on the planet"? I don't remember reading
about them entering the Loebner Prize... then again,
no, I don't think Mr. Mullins knows of the Prize,
either;

* So far chatterbots are concerned, the approach that
produced the best results was neglected for most of
its research history. Surprise, Mr. Mullins mentions
nothing of it;

* If it takes Cyc 3 million assertions to get close to
the "bare minimum" needed for it to achieve learning
independence, then, so far a certain quality level is
assured, size does matter. This is not addressed by
the story; it never mentions either that Alicebots,
too, live and die by the size almost as much as the
quality of their knowledge bases.

If this story was published on a non-scientific
periodical (like, say, Newsweek), I wouldn't be
bothered by the careless choice of cases, incomplete
display of evidence and overall sloppy research. The
fact that it was featured on New Scientist, however,
does make me to reavail the already limited amount of
trust I give the media.

-- 
Ja mata ne.
Helio Perroni Filho



	
	
		
Yahoo! Acesso Grátis - Internet rápida e grátis. 
Instale o discador agora! http://br.acesso.yahoo.com/