Modern corporations are rapidly finding themselves drawn into the world of Information Technology. Whether they venture into this realm boldly, driven by curiosity and innovation, or they are forced by the merits of their competition, businesses are constantly pushing the envelope of technological exploitation. The rate of this growth has been exponential throughout much of the 1990's and the advent of the 21st Century, culminating in an era where hardware and software advances have made machines of human likeness a very real and tangible phenomenon.
Artificial Intelligence, or AI, is now considered to be a critical part of the future of business. Expert Systems, Voice Recognition, Artificial Neural Networks, Interactive Marketing Tools, and Machine Vision are all aspects of AI that have made great strides in the past decade and have a promising role in the future of business.
As far back as the 5th Century B.C., Aristotle invented Syllogistic Logic, which is credited as the very first formal deductive reasoning system, a critical step towards the development of applications of intelligence and the same logic pursued by AI programmers today. In 1950, Isaac Asimov published his "Three Laws of Robotics", a statement of the ethics concerning AI development. In 1963, Thomas Evans' program, ANALOGY, demonstrated that computers are capable of solving the same analogy problems that are given on IQ tests. The 1980's saw the development and commercial use of expert systems, machines which could make intelligent decisions about particular processing functions. The most famous of these systems was Deep Blue, a creation of IBM, which in 1997 defeated World Chess Champion Gary Kasparov in a widely publicized match. Throughout the Nineties, AI manifested itself in increasingly sophisticated applications. The rapid advancement of AI during the latter half of the Twentieth Century was so expansive that today, at the beginning of the 21st Century, machines that can respond to visual and audio stimuli are quickly becoming commonplace. The World Wide Web has furthered the commercial interest of pursuing intelligent applications for increased consumer interest, reduced overhead, and ultimately higher profits. Today AI is evolving at a faster pace than ever before.
An expert system is one of the basic forms of artificial intelligence commonly employed today. These systems typically go beyond simple yes and no questions to arrive at answers with an accuracy that would generally be attributed to an experienced human. Expert systems are developed by interviewing human experts, divulging their decision making processes to pure logic, and then creating programmable rules that are based on the information obtained.
The core technology behind such a system is the inference engine. (Riley 1997) This allows the system to resolve the ruleset by which it achieves human-like results. Inference engines are often complicated programming routines that resolve interdependent IF-THEN statements forming the basic rules upon which the system operates. By definition, all expert systems use an inference engine.
The future of expert systems as a tool for Strategic Management is a promising one. Expert systems are already in use today to diagnose illnesses and assist in investment forecasting. (Kisly & Kisly 1999) An expert system is, by definition, a machine designed to take the place of or assist a human expert. They are designed by a knowledge engineer, someone who understands the way experts of a particular field think and arrive at conclusive decisions, and how to translate that process to terms a computer can understand. This typically involves making vast quantities of calculations, the core competency of machine technology, and then comparing the results of calculations against a scale of positive relevancy.
Such systems may find their way into the industrial automotive industry as formidable forecasters of demand. Geophysical and oilfield services may find expert systems useful in determining where to drill for oil. Expert systems could be employed by the military to determine probable outcomes of particular battle scenarios before they are fought, or perhaps even assist troops in the field with whatever regional geographic expertise they require. The possibility of expert systems for educational purposes is unlimited, finding practicality in virtually every field imaginable.
Expert systems, such as Deep Blue, still have a ways to go. Currently not even Deep Blue has the ability to "learn" new approaches to playing chess. (IBM, 1997) It can only evaluate possibilities (200 million possibilities a second), and make a selection based on rules and probability that were obtained from an expert (international chess grandmaster) and implemented by a knowledge engineer (IBM developer).
The true value of expert systems will be realized when they can communicate the logic behind their decisions to individuals in manner in which they can understand.
The marketing value of AI is quickly becoming realized by corporate entities throughout the United States and Europe. On the World Wide Web, new firms are finding value in having personalized, online marketing representatives to guide visitors through their site. These unique software products attempt to add a human touch to the website, displaying a character on the website who helps users navigate by communicating in an intelligible manner and responding to needs with maximum user friendliness.
The Alice foundation, founded by Dr. Richard S. Wallace, is perhaps the leading developer to the cause of online, Artificial Intelligence technology. The Alice project is an Open Source venture, meaning it is the end result of a world wide collaborative effort to create the most intelligible AI entity in the world. Its success has not gone unnoted, as it has won the Loebner Prize for Artificial Intelligence in 2000 and 2001.
Alice is a web entity that communicates with users by uniquely tying them to their particular IP address. The software stores information relative to the user of the IP address while communicating in a very informal manner. For instance, Alice may, in the course of conversation, come to know your name, which it will assign to a variable associated with your IP address. It may come to know what city you are from, and store this variable in memory as well. It may send a query to weather.com to see what the current conditions in your region are, and then make an intelligible comment based on that information. The results of this hierarchical technology are quite impressive to first time users, however they represent only the forefront of possible uses and exploits of such an application. Commercially, Alice and the AIML (Artificial Intelligence Markup Language) that she is based on, present the most extensible and versatile venue for modern advertising and marketing development. (Bush, 2001)
Alice, in fact, "learns" information about the various users that communicate with the program. Alice even communicates with other bots based on the same technology throughout the web, using a special protocol, the end result being a super-bot to one day be deployed which can operate with the combined knowledge of all the Alice programs running independently throughout the world.
Of course, there are problems associated with Alice even today. Alice cannot identify multiple users behind a proxied firewall or retain information about an individual if they switch IP addresses. Alice will assume that whoever accesses her from the same IP is the same person, which is not always correct. This obviously can lead to problems when the original user returns.
Still, businesses are finding increasing levels of success with technologies such as Alice and AIML. The core development team of the project continues to set admirable goals, including real time motion, audio interaction, and voice recognition.
Artificial Neural Networks are computing architectures loosely based on the learning design of the mammalian brain. In the human brain, various processing structures known as neurons communicate and change each other through connections known as synapses. An artificial neural net attempts to replicate this learning process by interconnecting a large number of processing elements (neurons) with mathematically weighted connections (synapses).
Learning within the human brain involves adjustments to the synaptic connections that exist between the neurons. Neural networks emulate this behavior as well, the weights of the connections storing knowledge necessary to solve problems. (Clark, 1997) These weights involve highly complicated mathematical algorithms, such as the graphical example posted below. (Stergiou, 1996)
Today neural networks are being applied to more and more real-world scenarios where pattern recognition is highly valued. Speech, signal, and character recognition are just a few of the roles that organizations have approached with neural networks. The possibilities for marketing and manufacturing firms to utilize dynamic intelligence to forecast demand and assist in strategic management could very well be a role that neural networks commonly fill in the future of commerce.
Neural networks seem to be capable of "learning", impervious to distortions within the input data, and often good at solving problems that people are talented at, but for which traditional technological methods are not.
Artificial intelligence has made tremendous advances in recent years. Learning models and expert systems posing an interesting future for businesses. Increasingly, businesses are relying on computers to take the role of human counterparts, not for their brawn, but for their brain.
Perhaps the most relevant test of success of todays time still revolves around the question that Alfred Turing posed when he challenged the world to develop a machine which would be indistinguishable from a human counterpart. Is it sufficient to defeat a world champion chess player for a machine to be truly intelligent? Is it sufficient to learn to recognize different patterns? To be able to see and hear, communicate, perhaps even with programmed wit, does that mean that a machine thinks? Perhaps machines will never be truly intelligent to the point that they replace their human counterparts. Kasparov may only be able to calculate three moves per second vs. 200 million moves per second by Deep Blue (IBM, 1997), yet his three moves are intuitive, guided by experience, and prompted by far more depth than any microprocessor can be programmed to obtain. Perhaps with the combination of the learning mechanisms of neural nets and the experienced decision making power of an expert system will the new level of Artificial Intelligence evolve.
The time is coming when thinking machines will be dispersed everywhere throughout business and the web, and the realization that our steps into this world described above were truly only the beginning.
Bush, Noel (2001). Why You Don't Need Proprietary Bot Software http://alicebot.org/articles/bush/dontpayalotforthatbot-00.html
Clark, David (1997). Introduction to Neural Nets http://members.home.net/neuralnet/introtonn/index.htm
International Business Machines (1997). Kasparov vs Deep Blue: a Contrast in Styles http://www.research.ibm.com/deepblue/meet/html/d.2.html
Kisly, Eugene and Victor (2000). Expert Systems http://www.wargaming.net/Programming/113/Expert_Systems_index.htm
Riley, Gary (1997). What are Expert Systems? http://www.ghg.net/clips/ExpertSystems.html
Stergiou, Chris (1996). What is a Neural Network? http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol1/cs11/article1.html