Artificial Intelligence (AI) conjures up visions of robots that can mix dry martinis while beating a grand master at chess; and to some, will one day be able to look, act, think and react just like a real person. I would like to explore the concept of AI as it relates to the business world, and its possible many other applications. I believe that true AI is a dream worth pursuing. Like me, there are many who, just like those of the early 1960’s, thought that putting a man on the moon seemed to be an extremely difficult, but not an impossible task, believing the achievement of true AI to come is just a matter of time.
To remain competitive, companies must continue to improve by doing better and doing more; all the while using fewer and fewer resources, especially, manpower. Greater numbers of the world’s companies are turning to systems, which they feel offer the best means of attaining these goals.
A group, or suite of tools that can help accomplish this pursuit of doing less with more is generally known as Decision Support Systems. This broad category usually consists of computer software and hardware, which includes Intelligent Decision Support Systems, Expert Systems and Artificial Intelligence. Do these systems really provide a valuable contribution to those who use them, and just how much faith can be put into them?
Strategic decision making concerns itself with determining where and how to deploy present resources to gain competitive advantages with the expectation of achieving some future reward. This simple, but powerful idea, permeates the planning process of large and small companies. Decisions related to how resources should be deployed consider specific measures necessary to compete effectively and efficiently; while strategic decisions are made with the expectation of improving future corporate profitability. Decision support systems are important additions in developing long term strategic plans, and thus long range profitability measures.
Before we can explore the possibilities and implications of AI, we must carefully define exactly what attributes make something “intelligent”. The most common way to define intelligence in through the term “consciousness”. A term such as this has no fixed definition; rather, it is a family of related concepts that tie together to form a picture of consciousness. Self-awareness, rationality, the ability for abstract thinking, and strategic thinking characterize consciousness. From this definition or description of intelligence we can gather that to exhibit true intelligence, there must be a conscious state, in other words, a state or condition of self-awareness. AI is broadly defined as anything that a computer does that we normally consider to be a human trait.
AI is the part of computer science concerned with designing intelligent computer systems, that is systems that exhibit the characteristics we associate with intelligence in human behavior—understanding language, learning, reasoning, solving problems and so on. Today’s AI sprung from the discipline commonly referred to Decision Support Systems, and as such, a true look at AI can not be conducted without first taking a look at its predecessors.
Why Have Decision Support Systems
Decision Support Systems provide a valuable data repository of lessons learned. By maintaining this data and providing real time updates, managers can help support their business choices by looking at a history of similar decisions made by others in their positions. This does not mean that every decision made, based on this data, will be good. However, it does help lower the probability of a bad one. This function alone can save a company from making a small error to making errors, which could threaten its ability to remain viable.
What types of Decision Support Systems are there?
Before we can understand the ramifications of these systems we must first explore the types and some of their features and functions. However, to understand them first we need to know what they are.
Intelligent Decision Support Systems:
This is a new paradigm for the DSS area. These extend the applicability and functionality beyond those traditionally covered by DSS applications and utilize a range of advanced technologies. The main role of an Intelligent Decision Support System in an organization is as an enabler for knowledge processing with communications capabilities. The approach, unlike traditional approaches in DSS, is that it does not focus merely on managerial decision-making, but attempts to reflect organizational realities. These systems usually consist of database which has a software interface designed to aid the researcher/decision maker with the information required to make an informed decision based on past events and experiences.
This is a research system that does just that, research. These systems use current information to make logical guesses and extrapolations about something unknown. These first appeared in the engineering field and other physical sciences, these computer systems dramatically decrease the time required to take a product or idea from concept to execution by running simulations within itself, locating problems, refining the model, and repeating these steps, gradually working the “bugs” out of the system.
Expert Systems are computer programs designed to review a set of facts (market conditions) and apply a set of rules (knowledge base) to arrive at the same conclusion that a team of experts would make if presented with the same facts.
Generally Artificial Intelligence (AI) is the discipline of building intelligence into computers. The term, AI, refers to a machine’s capability of processing data and responding with humanlike intelligence. They are essentially the Expert System taken to its next logical level of evolution.
“AI is having a Trojan renaissance,” says Nick Cassimatis, an AI researcher at the Massachusetts Institute of Technology’s (MIT) Media Lab, in Cambridge, Mass. Vendors are quietly building AI technologies into practical software applications that do everything from recommend music for Web shoppers to direct airplanes at airports. Because of AI’s tarnished reputation, vendors aren’t promoting their products as being AI-based. However, understanding the advanced AI technologies behind the products can help technology managers determine a product’s value and consider the potential of AI solving related business problems.
In business some of the most successful applications have been constructed by building substantial domain knowledge into computer programs. These systems are often referred to as knowledge base systems. Typically, these system use decision and process rules presented from experts to summarize that knowledge. Other systems use representations of cases from past experience to generate solutions for current situations, “case-based reasoning” (CBR). Law and other domains where reasoning is based on cases, find this approach very useful. Other approaches include so-called data mining and machine learning where knowledge is generated from an analysis of data. That knowledge is then summarized and used to make inferences.
Case-based reasoning is an approach to AI where a system stores case studies, responds to a problem by finding similar cases in its memory, and adapts the solution that worked in the past to the current situation. CBR sprang from cognitive science research, which was begun, in the early 1980’s by Roger Schank at Yale University’s AI lab, in New Haven, Conn.
Automated Customer-Support systems are an important business use of CBR. This is growing rapidly as companies look to reduce product support costs by encouraging customers to find their own answers on a web site instead, of calling expensive or toll free numbers.
An additional technology that has sprung from AI research and is finding a new home on the web, is rule-based expert systems. These systems, unlike collaborative filtering, typically use Boolean logic to process input from an individual user and employs stored rules to generate a prediction or suggestion. A prime example of this usage is the “Office Assistant” which is included with Microsoft’s Office 97 software package. This assistant is extremely useful for the individual who is unfamiliar with the software package. If the user seems to be floundering around looking for a way to accomplish a task, the assistant will attempt to interpret the desires of the user by looking at what he as been doing and then tries to make an educated guess as to what he wants to do. Then the assistant will display a help menu to guide the user through the desired course of action.
AI needs many ideas that have, up until now, been studied only by philosophers. This is because a robot or truly AI system, if it is to have human level intelligence and ability to learn from its experience, needs a general world view in which to organize facts. Others have pointed this out when addressing the necessity of broadening the professional constituency of AI and reexamining its fundamental assumptions about human nature.
One of the first successful applications of artificial intelligence in a business setting was the “Authorizer’s Assistant,” developed for American Express. The system allows the approval of most transactions without human intervention. Summarized in the system are a number of rules that relate to the approval of purchases. The system uses those rules and the unique profile that users establish by their pattern of purchases to ensure that the purchase is appropriate.
Perhaps the biggest return on AI is potentially on Wall Street. Substantial attention has been given to the development of automated trading systems, integrating AI into capital management, and using AI in capital planning. However, information about such systems is generally limited, since disclosure of successful approaches could lead to the loss of competitive advantage, and large sums of money. On activity that appears to be generating the greatest interest on Wall Street is that of data mining, using approaches such as neural networks.
Data Mining is the descendant, and to some, the heir and successor of statistics. Statistics and Data Mining pursue the same aim, which is to build compact and understandable models incorporating the relationships (“dependencies”) between the description of a situation and a result (or a judgement) concerning this description. The underlying assumption is that there is indeed some kind of dependency, i.e. the result, measurement or judgement we are trying to model is derived from some or all of the “descriptive variables” we have been able to gather. The main difference is that Data Mining techniques build the models automatically while classical statistics tools need to be wielded by a trained statistician with a good idea of what to look for.
Data Mining is the process of looking for knowledge and anticipating patterns in data. One of the primary approaches for finding patterns in data is neural networks. Neural networks were named, based on their structural similarity with the process used by the human brain. Although, the methods used by neural nets are beyond the scope of this paper, their applications are generally accessible. For example: a neural network approach can be used to investigate the relationship between a set of financial statement ratios and whether or not the firm goes bankrupt. Another example is for the case where banks must choose whether or not to make a loan, based on a set of input characteristics. In a similar manner, patterns of information are investigated using neural networks to assist in the process of choosing stocks as reported in U.S. News & World Reports.
So, we’ve explored what AI is and how it is being used today, but what about those dreams of a mechanical brain which so closely approximate the human mind that real life like robots are possible.
There is Cog, (Cognitive) which is the grand experiment in the latest approach to artificial intelligence: letting a machine discover the world on its own, the way humans do, rather than cramming its memory with some preexisting computer model that describes the world from a human perspective. Cog the android wannabe – wannabe because it doesn’t have legs yet. According to the creators, those will come later. For now, it’s still learning to coordinate its eye, head and hand “muscles”.
On the other side of the coin is Cyc (World Book Encyclopedia), the most ambitious version of the old school, top-down system. Some $40 million has been invested on organizing Cyc’s reasoning “engines” and stuffing its knowledge base with a half-million rules derived from 2 million common-sense facts. These are the things people soak up during childhood like: Mothers are always older than their daughters. Birds have feathers. When people that other software might miss, Cycorp has a database of captioned photos. Most database managers retrieve photos based on a precise word match in the caption. Type in “strong and daring person,” and Cyc pulls up a picture captioned “Man climbing mountain.” Cyc knows that a man is a person, and that mountain climbing demands strength and is dangerous.die, they stay dead.(World Book, 1999).
To show how Cyc’s common-sense method can help find information. The next stop for Cyc was to begin learning on its own by reading newspapers, books, and scientific journals. Then, in eight or nine years, Lenat figures Cyc will be smart enough for postgraduate work. It might help doctors make better diagnoses by checking medical records and presenting alternatives. Or it might help market researchers spot sales patterns missed by conventional data-mining programs.
Lenat expects Cyc to be ready to take charge of its own research lab by 2020. He expects Cyc to design unique experiments and uncover new knowledge. MIT’s Brooks has similar dreams for Cog’s offspring, but the timetable is less certain, because Cog got off to a later start. It was conceived just five years ago, after a Jan 12, 1992, part that Brooks gave to celebrate the birthday of HAL, the AI system in 2001: A Space Odyssey. After brooding about the lack of anything close to HAL, Brooks decided he had to take a stab at it.
If all goes well as more behaviors are added, such as a sense of touch and then smell, Brooks knows what he wants the results to be: something like Lt Commander Data, the super-smart android on Star Trek. How long might that take? Brooks doesn’t know. But maybe, around 2020, these two will mellow out and give us Commander Cycog.
What does this have to do with business? Well just think of the possibilities of a work force that never gets tired, requires little or no supervision and has the knowledge of the entire human race at its fingertips so to speak. The ramifications are staggering. This could be the only way that extended space travel may be undertaken. These AI Robots/Systems with virtual impunity could do extremely dangerous tasks, which would normally require a human to perform. Additionally, a work force of these machines could greatly increase production while lower overall cost of production. With no payroll, a company also doesn’t have to provide costly benefits. With the ability to learn these machines could be taught production changes in a fraction of the time required training a human workforce. Thereby reducing the time required to spin/tool up a new or modified production line, once again resulting in a cost saving to the company and ultimately the consumer.
Is artificial intelligence attainable? All the experts seem to think the answer to this question is a resounding Y E S. I agree with them, however, I don’t believe that the time lines they are forecasting are realistic. There are so many obstacles to overcome that I don’t believe 20 years will be enough time. Only time will tell if these individuals are on the right track. All we can do wait and see. However, it should be an exciting time for man and machine.
Computers and Internet