TITLE
    AI: Some Working Definitions Of The Terminology (Part 1 of 2)
Article ID:
Created:
Modified: 		3204
9/6/88
6/17/92
TOPIC

DISCUSSION



    As with any other computer field, "AI" (Artificial Intelligence) has a
    distinguishing set of buzzwords. Following are a few working definitions that
    either are used in this paper, or that one can expect to hear when talking with
    a typical "AI" professionals.


    Artificial Intelligence, or AI:

    Trying to define AI is similar to trying to define exactly what HyperCard is to
    someone who has never seen it (or worse, never worked with a Macintosh). The
    field of AI includes many areas, such as speech, vision, robotics, natural
    language interfaces, neural networks, machine learning, automated programming,
    intelligent agents and expert systems. Typically, when someone thinks of AI,
    they are associating a computer/robot with a more specifically human activity
    (thinking, seeing, hearing, understanding, moving).


    Common LISP:

    With the various versions of LISP available, Common LISP is recognized as the
    emerging industry standard for LISP today. This should help lead to a
    portability and consistency among architectures, allowing a complex problem to
    be developed with a LISP machine, while being delivered on a less expensive,
    possibly non-LISP architecture.


    Conventional vs. Symbolic Processing:

    Conventional software has typically been used for problems that are more
    algorithmic or numeric in nature and well understood, with the computer use
    being to perform very rapid calculations or to process very large amounts of
    data in an orderly fashion.

    Symbolic processing deals with information that is more conceptual ("symbolic")
    in nature, such as: "Given a certain CAD/CAM/CAE design criteria, what is the
    ideal design (lowest cost, most accurate, simplest to manufacture, most
    reliable)." While people will typically argue that you can do anything with any
    language (given enough memory, disk space, programmer time...), the main
    advantage of symbolic processing is to free the programmer/developer from
    dealing with the lower levels of implementation, allowing him to focus on a
    higher, conceptual level of abstraction to address his real problem at hand.

    Another way to think of symbolic processing is the concept of symbols.
    Typically a symbol represents more than just physical definition. If someone
    asks for a "pen", they are probably not as interested in the physical features
    (gold cross, BIC, apple logo or size = 5.75 inches) but rather are asking for
    something that will write in ink.


    Data Type Processing:

    LISP programs typically use a "tag" implementation that allows data types to be
    checked at run-time, as opposed to mandating that data types be declared in
    advance. This gives applications more flexibility to deal with complex and
    changing situations over the development life cycle.


    Development vs. Delivery machines:

    This distinction indicates the difference between the computer system on which
    an application is developed , and the computer on which the final application
    is delivered to the end user.

    An example: a company might use a LISP machine and AI software package (which
    may total well over $100,000 for initial hardware/software costs) to develop a
    solution, but deliver it on a microExplorer or Mac II for a much lower cost per
    end-user. The idea is to have the ideal development environment to save
    development cost and increase programmer productivity, while also being able to
    deliver the solution in a relatively inexpensive environment.


    Domain Expert:

    One typically recognized as an expert in a field (domain).


    This article is continued in "AI: Some Working Definitions Of The Terminology
    (Part 2 of 2)".



Document Information
Product Area: Computers
Category: Troubleshooting
Sub Category: General Topics

Copyright © 2000 Apple Computer, Inc. All rights reserved.