|
|
|
One of the best modern definitions is the one found in the Microsoft Press Computer Dictionary: "The branch of computer science concerned with enabling computers to simulate such aspects of human intelligence as speech recognition, deduction, inference, creative response, the ability to learn from experience, and the ability to make inferences given incomplete information." The definition goes on to say: "Two common areas of artificial-intelligence research are expert systems and natural-language processing."
In the upcoming Part 2 of this discussion on AI we'll take a look at some of the more serious and ambitious definitions of AI research and applications. However, for now, the above is as good a definition as any to describe the majority of AI technologies that are being put to use in business applications.
"Expert Systems", also known as "Rules-Based" systems, are an attempt to put into computer form the expert knowledge of one or more people from a particular occupation and/or field so that the computer can perform that specific task or tasks. That's why you'll notice that the tasks listed within the above definition, wide-ranging though some may be, basically remain confined to one specific area: speech recognition, natural-language recognition, etc.
Because talking, thinking computers like Star Trek's "Data" and others that can do almost anything a human worker can do are still the stuff of science fiction, most AI efforts revolve around the more realistic possibilities of computerizing specific jobs that are well suited to the process.
Computers have always been best at repetitive tasks such as crunching numbers simply because they can do the job thousands if not millions of times faster than a human worker. In contrast, computers have always performed poorly at tasks that require subjective reasoning and thought.
For example, even bargain-basement-priced PCs can be programmed to "analyze" puzzle games. I'm talking here about the kinds of puzzles that require you to assemble the pieces into a picture or a shape, such as a castle, etc. By analysis I mean the computer can easily count the number of pieces to tell you how many there are, and do such things as tell you how many possible combinations and ways of solving the puzzle there are, etc. For the computer to actually solve the puzzle, however, is an entirely different matter. That process requires "intelligence."
Even so, it is entirely possible to program a computer using expert system methods so that it contains the same kind of knowledge and rules a person would apply to solving the same puzzle. From then on the computer would be able to solve the puzzle faster than any human, regardless of how scrambled and scattered the pieces were to begin with. Notice I said "the" because expert systems are usually built to solve specific types of problems or perform specific types of tasks. So, the above examples refer to a program designed to solve only one specific puzzle (a picture of a horse, for instance).
It is possible, but considerably more complicated, to design an AI-based program capable of solving more than one puzzle of the same type. So, for instance, even though one puzzle might represent a picture of a barn, and another a picture of a house, and another a picture of an airport, sufficient "intelligence" could be programmed in so that the computer could solve most puzzles of that type, regardless of the content.
How does this relate to the workplace? Well, a large percentage of business processes are similar in scope and complexity to the puzzle examples above. Meaning that while currently only a small percentage of the actual work is being done by computer and the rest by your staff, such processes are candidates for AI-based automation. Using the methods discussed above you can automate those portions of the process that currently require the greatest investment in time, thus saving both time and money.
Incidentally, some of the most common AI applications that aren't confined to the laboratory
are:
 | The voice recognition used by some phone systems ("Touch or say 1, now," etc.), as well as the voice recognition software that is now readily available for the PC.
|
| The OCR (Optical Character Recognition) software long used by banks to read your deposit slips, as well as the many other business forms and applications designed for computerized processing. And, of course, OCR software is what allows scanners to create not just images (photocopies) of printed pages, but to "read" the text as well. |
Perhaps the closest partnership between AI technology and a traditional, and highly successful, business application is the AutoLISP implementation of the AI programming language LISP, which is found in the popular AutoCAD drafting program. (Lisp is the most popular of the AI programming languages, and, even though more and more business applications are being developed in Lisp, it is still more commonly used for research-oriented projects.)
Send mail to with questions or
comments about this web site.
Copyright © 1998-2003 GRPMAX. All Rights Reserved. Marks used are the
trademarks and/or registered trademarks of their respective owners, including
GRPMAX and Phillip A. Covington.
|
|
|
