next up previous contents index
Next: Related Work Up: Research Topics Previous: Program Synthesis

Artificial Intelligence

  The language of Nuprl has been described as a ``logic for artificial intelligence'' [Turner 84]. The system can be viewed as an intelligent system , and many of the things we do with it and to it might be characterized as improving its ``intelligence''. As we develop more ``expert  reasoners'' similar to those for equality and arithmetic, for example, we certainly increase the deductive power of the system, especially as we allow them to cooperate [Nelson & Oppen 79]. The natural growth path for a system like

tends toward increased ``intelligence''. This is obvious when we think about adding definitions, theorems and methods to the library. It is less obvious and more noteworthy that other changes move it in this direction as well. For example, it is helpful if the system is aware of what is in the library and what users are doing with it. It is good if the user knows when to invoke certain tactics, but once we see a pattern to this activity, it is easy and natural to inform the system about it. Hence there is an impetus to give the system more knowledge about itself. This is related to recent themes in the field of artificial intelligence [Bundy 83,Charniak & McDermott 85,Davis & Lenat 82].

Richard Eaton
Thu Sep 14 08:45:18 EDT 1995