From: yaoziyuan@gmail.com   
      
   Introducing "Manned Intelligence"   
      
   I have long felt that AI should have some human ingredients to be   
   practically powerful. For example, pure AI programming languages like   
   Prolog and Lisp may require too much labor to formally model a problem   
   that involves commonsense knowledge. I think a "program" can have some   
   parts executed by a real human (so that it can exploit commonsense   
   knowledge stored in the human's mind). That actually happens every   
   day: a "program manager" is someone who writes "human programs"   
   executed by human employees.   
      
   I envision that the computer can serve as a "strategic adviser" that   
   presents "general problem-solving strategies" when a human tackles a   
   problem. This is like a modernized version of George Polya's book "How   
   to Solve It". In the very beginning of a problem-solving process, the   
   computer can give these "First Aid" tips to the human:   
      
   * Can you google for a solution?   
   * Can you ask friends online for a solution?   
   * Can you consult Wikipedia for a solution?   
   * Can you go outside and ask for help?   
   ...   
      
   Sometimes we have epic failures at an easy problem simply because our   
   minds aren't right at the moment and we forget to ask for help (or our   
   minds don't think anyone can help). If we strictly carry out a   
   flowchart like the above, we may overcome our minds' temporary   
   "downtime".   
      
   AI since its birth till today has been mainly about "unmanned   
   intelligence", whether it be knowledge-based or statistical. Compare   
   it with the history of human aviation: first we had manned aircraft,   
   and only recently did unmanned, autonomous aircraft like the X-47B   
   came into being. That's the right, natural way to develop things. Can   
   you imagine the Wright brothers predicting autonomous starships after   
   their first successful flight? Unmanned intelligence, like unmanned   
   aviation, can't become mature until manned intelligence is fully   
   explored.   
      
   Peter's AIMA book is a good account on unmanned intelligence, but I   
   haven't seen any serious book or academic program for manned   
   intelligence. The field hasn't been created yet.   
      
   So what is manned intelligence? Generally it's about shifting AI   
   subtasks that are hard for the computer but easy for the human to the   
   human. Specifically, commonsense knowledge representation and   
   reasoning are trivial for the human but hard to codify and execute   
   with the computer. So I imagine, in a problem-solving process, the   
   computer provides high-level, strategic guidance, while the human   
   takes care of low-level problem-solving. The "First Aid" example I   
   made above is a good example of very high-level strategic guidance.   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)