The next logical revolution occurs at some time in the future when we get our act together on the exploitation of logic in IT and succeed in realising intelligent systems which are better at programming than human beings.

• Automated Intelligent Problem Solving
  • including writing programs
    Teaching machines to write programs is a milestone in the development of artificial intelligence. Once they can do that they are much better placed to learn new methods and go beyond our own intellectual capabilities (in relation at least to logical problems).
• Logic for interface specifications
  The first stage is a simple extension to the languages we use for programming, so that there is room for the kind of precise semantic specification which would enable computers to use the functions thus specified.
  • think of as an extension to IDL
    OMG introduced IDL (Interface Definition Language) for describing interfaces, but omitted to allow any way of describing the semantics of the interfaces. An extension would allow the semantics to be specified (not mandate it), which would provide a basis for re-use by machine rather than man.
• Logic for data/knowledge bases
  • providing a framework for deductive problem solving
    Query languages naturally already express the semantics of a query, the key here is to integrate the languages used for interface specification, for expressing queries and for storing information.
• The Singularity
  Infinite intelligence out there on the net, resulting from runaway escalation once the first superintelligent AI is established.
  • AI => AI+ => AI++=> AI+++=>....
    Once we have an AI, it can design a better one, and the better one can design an even better one...
  • each iteration takes less time than the previous one
    This creates a runaway sequence which theoretically would yield infinitely intelligent artifacts in a finite time. Many people think this will happen within the next half century.

---

© created 97/9/18 modified 97/9/27