[extropy-chat] Twenty Questions Twenty Questions -Theneural-net on the Internet

[spike]

> bounces at lists.extropy.org] On Behalf Of Erik Starck
> Subject: RE: [extropy-chat] Twenty Questions Twenty Questions -Theneural-
> net on the Internet
> 
> At 03:45 2005-02-19 spike wrote:

> >...  Friends, we
> >have survived to see computers pass Turing's criterion.
> 
> Hardly. This 20q game is a constrained environment, not that different
> from
> dynamically building SQL queries to a database. It's a cool technology
> demo
> and fascinating display of how a relatively simple program can learn from
> its' users and improve it's own capabilities, but it's not closer to
> passing the Turing test than say for example the Amazon suggestion
> database. Erik


What we have learned is that the Turing test can be challenged
with nothing more sophisticated than a giant lookup table.
I have long suspected that much of human-speak is nothing
more than a de-facto lookup table.  If we could get a ton
of data from teen chat sites, randomly generate comments
such as "sup", "how r u?", "shizzle be down widdat, dawg," 
all liberally sprinkled with the universal functional adverb 
"like", we could create a lookup table that would do an 
adequate job of simulating an unsophisticated human, 
possibly good enough to fool some on Turing's criterion.

spike