The discussion of artificial intelligence may seem too “sci-fi” and unbelievable for some.  It is difficult to believe that someday in the near future, the human race may actually “give birth” to a different kind of intelligent organism.  To make this discussion more concrete and credible, we would like to offer an example of an artificial intelligence system that is capable of learning.  This system, developed by Intelligenesis Corporation, is an intranet-based knowledge management system.  Its uses thus far have been limited to the areas of business and financial markets, but its possibilities are worth discussion.

            Intelligenesis calls their system Mebmind.  From the users’ perspective, Webmind is a system for posing and answering questions.  Internally, however, the structure and processes for answering and learning are much more complex.  The goal of a Webmind program is to understand data stored on computer networks, create its own information using this data as a starting point, and answer questions regarding this data.

           For example, Webmind has already demonstrated its usefulness in the financial services sector.  It is capable of predicting trends, addressing conceptual issues of market directions and policies, and automatically building and evaluating trading systems.  How, you might be asking, can a computer forecast trends in financial markets that have proven very random and unpredictable in the past?  According to information found at Intelligenesis’s website, “financial markets are complex and difficult to predict in general, but there exists windows of enhanced predictability.  Webmind widens and makes more certain the boundaries of those windows of enhanced predictability, as well as providing general predictive improvement.”

            How does all this occur?  It would take a book to dissect the processes in detail, so we will attempt a rough outline.  To begin with, just as humans need eyes to perceive sights and ears to perceive sounds, Mebmind must be given perceptual “methods” for processing various types of information.  Webmind contains a self-organizing network of information-carrying agents called the “Psynet.”  The data stored in the Psynet is allowed to discover its own structure rather than having structure imposed on it.

The information-carrying agents take on three distinct forms:  static, relational, and mobile.  Static agents are also called “nodes,” and they are static in the sense that they have a continued existence in the Psynet.  Relational agents are also referred to as “links” that are able to connect the internal structure of the Psynet with the external structure of other networks, the internet, and intranets.  Finally, mobile agents change with time and they are responsible for the “learning” of relationships. 

Learning takes place in several ways.  The Psynet and its agents are able to recognize patterns in data, recognize relationships between nodes, and they are able to create new nodes that represent the formation of “concepts.”  The Psynet is also able to learn through direct introspection wherein it poses queries to itself.  They argue, however, that greater intelligence will be achieved by networking Psynets together and allowing them to interact and query each other.  The ability of an organism to learn and evolve is one of the fundamental characteristics contained in the theory of strong artificial intelligence. 

The creators of Webmind use their program as the foundation for their discussion of a “World Wide Brain.”  They propose that in the not too distant future, their technology will evolve allowing many large intelligent networks to interact with each other producing an intelligent and self-aware system.  This hypothesis resembles the fictitious Skynet system seen in the movie Terminator 2.  Check out our discussion of the movie to learn how this movie views the potential disasters that can arise out of such a system.

This project was produced for PSY 380, Social Psychology of Cyberspace, Spring 2000,  at Miami University. All graphics in these pages are used with permission or under fair use guidelines, are in the public domain,  or were created by the authors.  Last revised:  Monday, April 15, 2002 at 22:03:34.  This document has been accessed  2  times since 1 May 2000.  Comments & Questions to R. Sherman