Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech2!pirates!cssun.mathcs.emory.edu!swrinde!ihnp4.ucsd.edu!newshub.nosc.mil!news!nebiker
From: nebiker@nosc.mil (Ralph R. Nebiker)
Subject: Re: AI and the future
Message-ID: <1995Nov30.212138.22512@nosc.mil>
Sender: news@nosc.mil
Organization: NCCOSC RDT&E Division, San Diego, CA
References: <vcd103.5.00585A28@psu.edu>
Date: Thu, 30 Nov 1995 21:21:38 GMT
Lines: 26

vcd103@psu.edu (Vinay Desai) writes:

>I had an interesting discussion with a few colleagues about AI recently.  
>During that conversation, someone compared the present euphoria over the WWW 
>to that around AI back in the 1960s.  Is this a fair comparison and if so, has 
>AI delivered all that was promised?  Comments?

>VCD

   For all practical purposes, I think so.  The latest example which comes to
   mind is natural language processing.  It turns out that our human facility
   for language is more an instinctive, born-with-it capability than it is a
   learned capability.  Language has been the primary vehicle for assessing
   how we learn and process "data".  It turns out we know very little about
   the acutal learning process, but we have accumulated an awful lot of
   data describing what we believe are symptoms and indicators of the
   learning process.
   
   At this stage it seems to me we are feeling around in the dark, hoping
   we grab onto something that will point the way; even if we had no idea
   what we grabbed onto.  We have much farther to go in explaining intell-
   igence, much less replicating its function, than we had previously thought.
   
   It seems to me expert systems will be the big productivity item for years
   to come.
   i
