ARTIFICIAL INTELLIGENCE

Artificial Intelligence



Artificial Intelligence

Introduction

Artificial intelligence (AI) is the study of how to build intelligent systems. Specifically, AI is concerned with developing computer programs with intelligent behaviors, such as problem solving, reasoning, and learning. Since the name artificial intelligence was coined in 1956, AI has been through cycles of ups and downs. From the beginning, game playing has been an important AI research subject for problem solving (AAAI, 2006). In 1997, IBM's Deep Blue chess system defeated world-champion Garry Kasparov. Expert systems have been successfully developed to achieve expert-level performance in real-world applications by capturing domain-specific knowledge from human experts in knowledge bases for machine reasoning, and they have been used widely in science, medicine, business, and education (Clark, 2004). Machine learning arose as an important research subject in AI, enabling dynamic intelligent systems with learning ability to be built.

Patterned after biological evolution, genetic algorithm spawns the population of competing candidate solutions and drives them to evolve ever better solutions. Agent-based models demonstrate that globally intelligent behavior can arise from the interaction of relatively simple structures. Based on the interaction between individual agents, intelligence is seen as emerging from society and not just as a property of an individual agent (Forest, 2004). Recently, AI has been moving toward building intelligent agents in real environments, such as Internet auctions. Multi-agent systems, as the platform for the convergence of various AI technologies, benefit from the abstractions of human society in an environment that contains multiple agents with different capabilities, goals, and beliefs.

Usage of AI in Different Fields

The AI approach has been applied to social, psychological, cognitive, and language phenomena. At the theory-building level, AI techniques have been used for theory formalization and simulation of societies and organizations (Georges, 2003). For example, a GAME THEORY problem, the “prisoner's dilemma,” is a classic research problem that benefits from computer simulation in exploring the conditions under which cooperation may arise between self-interested actors who are motivated to violate agreements to cooperate at least in the short term. More recently, the multiagent system emerged as a new approach of modeling social life (Perkowitz, 2005). These models show how patterns of diffusion of information, emergence of norms, coordination of conventions, and participation in collective action can be seen as emerging from interactions among adaptive agents who influence each other in response to the influence they receive.

At a data analysis level, AI techniques have been used to intelligently analyze qualitative data (data collected using methods such as ethnography) using symbolic processors and quantitative data using AI enhanced statistical procedures (Holland, 2005). For example, it has been shown that neural networks can readily substitute for and even outperform multiple regression and other multivariate techniques. More recently, the process of knowledge discovery has shown great potential in social science research by taking the results from data mining (the process of extracting trends or patterns from data) and transforming them into useful and interpretable knowledge through the use of AI techniques.

Medical

With the introduction of complex medical expert systems, it is becoming increasingly likely that in the future we will ...