From: abucci@cs.brandeis.edu   
      
   > If a huge company like Microsoft would concentrate on just one thing -   
   > building an AI system - they could do it.   
      
   Hmm, I find that doubtful.   
      
   > Microsoft is one of the few companies who would have the power to do it.   
      
   Do you lack trust in the academic establishment?  I'm pretty sure the key   
   insights that lead to AI, whenever it happens, will come from   
   universities, not from companies.   
      
   > Birds and planes rely on the aerodynamic principle of lift.   
   > It was important to discover the aerodynamics principles, and   
   > with the power of oil engines we achieved the ability to fly.   
   > In AI, it is important to discover the fundamental principles, and   
   > with the power of computers, networks and the internet,   
   > systems will probably achieve the ability to "think".   
      
   I detect a common misunderstanding here.  The principles of lift were   
   known well before the Wright brothers.  There were gliders, for instance.   
   Further, thrust was not the problem either -- people had tried combustion   
   engines on flying machines.  The key insight the Wright brothers added was   
   how to control the craft once it was flying.  That's where everyone else   
   failed and they succeeded.   
      
   This was a scaling problem -- how to maintain control over the craft as   
   its weight and airspeed increased.  The crucial invention was the aileron.   
   The reason is, it permitted a heavy, fast-moving craft to be controllable   
   by a relatively weak and light human being.  Gliders were controlled by   
   the human operator shifting their weight left and right to turn.  With a   
   heavy fast-moving craft with high frictional forces, shifting weight no   
   longer sufficed for turning.  That's where the aileron was a brilliant   
   addition.  By bending the wing, the control power scaled with the speed of   
   the craft, exactly as you need it to for controlled flight.   
      
   The analogy with AI is interesting.  You seem to assert it was the   
   powerful engine which permitted flight.  Similarly, people have thought   
   that scaling computers to very large sizes would permit AI.  Some people   
   even think there will be a magical transition from not-smart to smart,   
   simply because of scale!  However, I feel in both cases there is   
   fundamental misunderstanding and error in that point of view.  Principles   
   of lift were already known before the Wrights, people had already tried   
   combustion engines to make airplanes, and they failed.  Similarly, we have   
   extremely large computers, but they are not smart.  It seems pretty clear   
   that if we scale the computers even larger, they still will not be smart.   
      
   Couldn't it be we lack a principle of control in our computers?  The   
   computational equivalent of the aileron?  It's possible we're in the same   
   state the Wrights were in.  We have powerful computers, we have a lot of   
   know-how, but we're lacking something basic about how to control them.   
      
   Anthony   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)