Artificial intelligence is an area that has not fully matured. Since John McCarthy first mentioned this term in 1956, countless applications of this new tool have been thought of. Applications ranging from the trivial to the formal, from games to health, from war to saving lives. But there are countless areas that have not yet benefited from the benefits of this technology, or have not used it on a large scale. Specifically, the business area is one of them, which offers a very wide and attractive range that can be easily filled by Artificial Intelligence, helping in the creation of expert systems or functioning as a component of more complex decision-making systems, to be used as invaluable tools by current administrators.
Introduction
The use of information technologies in daily life has proven to be a tool that offers infinite competitive and resource-saving advantages that were previously unavailable. In addition, these technologies allow the generation of areas of opportunity, which, when used correctly, offer great benefits to all those who apply them. In one of these niches of opportunity are artificial intelligence systems, which have been applied in a wide variety of tasks, from teaching to the automation of production processes.
Within the great panorama of opportunities that we have to be able to apply artificial intelligence in companies, the presence of a special point is very noticeable, “Artificial intelligence in business”.
On many occasions, the core competence of a business lies in the correct decision-making, in a timely manner. But it is also difficult to find people who have the necessary characteristics to make the right decisions at all times. These administrators need tools that allow them to visualize in a general way, the strengths and weaknesses of their business, as well as the areas of opportunity that could arise. They also need systems that help them improve the quality of the decisions that are made every day.
Most artificial intelligence systems have the peculiarity of "learning", which allows them to improve their performance as time passes. In addition, these systems can analyze very large volumes of information at very high speed, which allows obtaining specific indicators of the company's operations.
In this article, some of the most important characteristics of the use of Artificial Intelligence (AI) in business will be exposed, its components, areas of opportunity and weaknesses, as well as the variants that can be derived from technology and the way in which they can be applied in the real world.
Methodology
The methodology used was a search for information on the Internet, on sites dedicated to artificial intelligence, supported and reinforced with first-hand information obtained from searches in the digital library of the Instituto Tecnológico y de Estudios Superiores de Monterrey (ITESM). The databases consulted in this library were: ACM Digital Library, IEEE Digital Library and Proquest Computing.
Background
In fact, the idea of looking for principles to govern intelligence and in turn incorporating them into a machine is very old. There are indications that some old civilizations, such as the Greek, the Chinese and even the Maya, had already formulated some mechanisms, sometimes real or fictitious, for this purpose.
Later there were other aspects such as Aristotle and his syllogisms. But really the interesting thing begins in the time of Leibnitz, who sought to find an algebra capable of representing all knowledge, including moral and metaphysical truths, to achieve a deductive system. He managed to build an abstract system called philosophical or reasoning calculus, but he did not have the technology to implement those ideas. (Doyle 1996)
In 1849, George Boole developed the principles of propositional logic with which he intended to collect some fundamentals related to the nature and constitution of the human mind. He wrote a work called "An investigation on the laws of thought, on which the mathematical theories of logic and probability are based." He was the creator of the truth table, which even now is widely used in the entire mathematical and computational field. (Fig 1.)
Fig 1. Boolean truth table
Some time later in 1874, Frege proposed a notation system for mechanical reasoning, which is the antecedent of predicate calculus, which he called "concept writing." It took almost a century for Turing to propose the first modern article on the possibility of mechanizing intelligence in 1950.
The introducer of the term "Artificial Intelligence" was John McCarthy in 1956. It was taken from the name of a conference that was formed thanks to the appearance of a book, written by him in conjunction with Claude Shannon (1956), dealing with the Fundamentals mathematicians of the theory of automata, oriented to the formalization of ideas about intelligence. Later, Allen Newell, together with Herbert Simon and Cliff Shaw, developed a program capable of proving theorems of propositional logic. This program was later called GPS (General Problem Solver). All these works contributed to the development of Artificial Intelligence, considering that the heuristic search was one of the fundamental methods to solve problems.Another important tool was the system of physical symbols to represent problems and their solutions (García 2002).
The person who is recognized as the father of Artificial Intelligence is Herbert Simon, along with Allen Newell and Marvin Minsky. Simon proposed the physical symbol hypothesis as a way to represent and model human thought.
On this line, the development of Artificial Intelligence was followed in its first stage. Unfortunately after this time of great growth, the expectations were very high. Very ambitious projections began to be made, which due to the technological level of those times were impossible to implement (García 2002).
From various points of view, especially that of business, artificial intelligence was expected to offer more and better results because there were expert systems capable of emulating the reasoning of experts to diagnose or predict, perhaps even plan. But it is also true that progress at this stage was restricted by the inability of systems to adapt to uncertain and imprecise environments.
In 1984, E. Dyson predicted that AI was not mature and therefore had no commercial value. It should be captured in strategically important systems to build your credibility. This prediction held true until the mid-1990s, which is why many people considered AI to be dead. By this time some lessons had been learned such as:
- Recognize that computer and AI do not replace human intelligence, but rather complete it. Accept that the most important capital in a company are the people, the experience and the knowledge that has been accumulated over the years. Smart systems will help improve personal computing. The use of AI in business should be focused on solving customer problems.
It can be deduced that in this first era of "Modern" Artificial Intelligence, many challenges were presented, because in order to achieve the high expectations that had been forged, the appropriate technology was not available and, on occasions, neither the in-depth knowledge of the themes. Both technological and theoretical development was lacking. In addition, many people did not bet on this new concept.
Fortunately, as we will see later, a new and important turn has been given to AI, based on new approaches and new ideas, which herald a promising future.
The resurgence of Artificial Intelligence
In recent years, there has been a significant rebound in AI. There are many characteristics that make this tool once again show excellent conditions of being an applicable technology, which will help companies be more productive. One of the new approaches is the so-called “behavior-based”. This approach provides a broad idea of the general concept of Intelligence, assuming that it depends entirely on the way in which the behavior of a system contributes to carrying out its functions. There are also new techniques that can be applied in this new AI, such as neural networks (Fig. 2), genetic algorithms, Bayesian networks, learning, etc.
Fig. 2. Representation of a neural network and a simple neuron.
Genetic algorithms is one of the new techniques that has received the most support and promises the most future. It is interesting to mention that they are completely based on and modeled on natural selection processes that have been identified in living organisms. Just as in nature, species mutate according to the characteristics of the environment, time or other external factors, the systems that handle this concept have the ability to adapt responses to changing problems according to time. This is that the behavior of these systems is mutated according to the characteristics of the environment and the problem.
These types of algorithms have a huge advantage over traditional AI symbolic systems. The latter are too static and not prepared to withstand sublime changes in problem conditions. Usually, symbolic systems are developed to solve only one type of problem. If this problem changes in any condition, you do not have the opportunity to change to try to solve it. For this reason, systems that operate with genetic algorithms have a further future in this field.
Now the expectations of Artificial Intelligence are based on the fact of seeing it as an investment option in the development of tools that reduce costs, distribute resources, detect fraud, online help, capture knowledge, etc.
But artificial intelligence, in addition to being able to be applied to different production processes (which can range from food generation to weapons construction), can also be applied to different areas of opportunity in the economy, finance, or for decision-making in companies. corporations.
Where, when and how
With all this new potential injected into Artificial Intelligence, and based on its origins, it is easy to imagine the wide range of opportunities where it could compete using it. As an example you can say:
Where?
In management areas of small and medium-sized companies.
When?
When the size of the company prevents having people of high level and high knowledge for decision making. Sometimes a vicious circle is formed that is generated in the following way: The company does not have growth because the correct decisions are not made. You cannot hire a person who has the decision-making capacity to make the company grow, since as it has no growth, they cannot be hired.
How?
Implementing or subletting an expert system that has the facility to help current directors and administrators to make correct decisions, by analyzing the risks and advantages of different opportunities and offering a viable alternative for growth.
Another example could be:
Where?
Banks, exchange houses, credit card companies.
When?
When you want to improve anti-fraud systems.
How?
Using neural networks to study credit usage patterns and detect potentially fraudulent transactions. According to Widrow, Rumelhart & Lehr (1994), credit card fraud is a growing problem that affects the entire industry.
Finally, you can have this example:
Where?
Financial institutions and brokerage houses
When?
When you want to have a higher return is the transactions and a lower degree of risk when investing.
How?
Expert systems could be used that use neural networks that learn from genetic algorithms, that learn and train so that they can make financial predictions. In addition to the speed and processing power currently offered, this could be done for many companies with different features at the same time. This is currently being piloted by some financial companies, such as Merrill Lynch & Co., Solomon Brothers, etc. (Willow, 1994)
In addition to these examples, one can think of the following recommendations (García 2002), which could be sources of interest for the application of artificial intelligence:
- Use of virtual learning to train people. Resolution of updating problems. Information and knowledge conservation activities. Development of emotive computing and its use to generate better interfaces. As a support to recovery from illnesses (biomedicine). Help to enable the use of devices and equipment intended for people with different abilities.
Conclusion:
During the information gathering process, the interest and expectation generated by the use of technology and especially artificial intelligence in business has been denoted. There are still many things to define, conceptualize and accomplish. But it is also true that business conditions have drastically changed in order to adopt information technologies in a natural and friendlier way. Apparently, they are no longer seen as a latent threat, but are taken as tools to enhance the performance and performance of each of the members of an organization.
Returning to the subject that concerns us, it is pleasant to see that models that use artificial intelligence for decision-making are increasingly used and valued. Seeing this behavior, it is easy for us to ask ourselves: What will the future hold? How will these types of tools be developed in the short, medium and long term? Just thinking about the answer again presents us with a very broad picture where staggering amounts of opportunity areas are in sight waiting to be taken.
Bibliography
Biasca, Rodolfo “Performance Management, the 10 steps to build it
Burghart, JH (October 1998) "Design of a course in artificial intelligence and expert systems" IEE digital Library. Proceedings of the Frontiers in Education Conference. (Consulted on May 9, 2005).
Chan, Nicholas. "Information Dissemination and Aggregation in Asset Markets with Simple Intelligent Traders", MIT, September 1998, Massachusetts Institute of technology
Dean, Thomas (December 1996) "Strategic Directions on Artificial Intelligence" ACM Computing Surveys, ACM Digital Library. (Consulted on May 12, 2005).
García Vega, Angelica "Artificial intelligence in business". February 2002, University of Veracruz
Hightower, James K. (1985) "Tutorial on artificial intelligence". ACM Digital Library
. Proceedings of the 17th conference on winter simulation. (Consulted on May 12, 2005)
Kingsbery, James. "Economics and Artificial Intelligence" Economics and Artificial Intelligence, April 2005, AI depot
Kumar, Vinay. "Learning based approach to estimation of morphable model parameters", MIT, September 2000, Massachusetts Institute of technology
Lane, Kenneth M. (March 1995) “Neural networks from idea to implementation”. ACM Digital. (Consulted on May 12, 2005).
Markland, Robert. (December 1990). "Artificial Intelligence, Heuristic Frameworks an Tabu's Search" Managerial and decision economics. Proquest, (Accessed on May 10, 2005).
Minsky, Marvin. "Linear Decision and Learning Models", MIT, March 1968, Massachusetts Institute of technology
Punt, Michael (November 2002) "A taxi ride to late capitalism: Hypercapitalism, Imagination and articial intelligence" ACM Digital Library. (Consulted on May 12, 2005).
Reese Hedberg, Sara (August 1996) "AI Toold for Business Process Modeling". Executive Inside. IEEE digital Library, (Consulted on May 08, 2004)
Roque Moranchel, Silvia "Business intelligence". March 2001
Sviokla, Jhon (October 1986) "Business Implications of knowledge based systems" ACM Digital Library. (Consulted on May 09, 2005).
Tafti, Mohammed (1990) "Trends and Direction in Experts Systems". ACM Digital Library. Proceedings of the 1990 ACM SIGBDP conference on Trends and directions in experts systems. (Consulted on May 12, 2005).
Wellman, Michael P. (December 1996) "The economic approach to artificial intelligence" ACM Computing Surveys. ACM Digital Library. (Consulted May 12, 2005).
Widrow, Bernard & Rumelhart, David E. & Lehr Michael A. (March 1994) "Neural Networks: applications in industry, business and science." ACM Digital Library. (Consulted on May 12, 2005).
Wong, Kit Po (December 1993), “Artificial intelligence and neural network applications in power systems”, IEEE digital Library, (Accessed May 08, 2004)
Wu, Xindong (September 2004). "Data mining: artificial intelligence in data analysis". IEEE digital Library. Proceedings of the International Conference on Intelligent Agent Technology 2004. (Accessed May 8, 2004)
