Applications of Artificial Intelligence
Artificial Intelligence (AI) has always been at the cutting edge of the development of Computer Science. It has also inevitably played a fundamental role in realizing the most advanced application of computers, even if the debt to AI has occasionally been masked by the rapid adoption of AI concepts and techniques into mainstream AI. Indeed, John McCarthy famously remarked in the early 1960s that AI is tomorrow's computer science. This Special Issue of Artificial Intelligence exemplifies the role of AI in an increasing diversity of applications: from a system to teach thermodynamics to novel software paradigms that realize challenging business goals; from robot tour guides to a system to discover protein structures in oncology research, and from a space shuttle mission planning system to one that supports the diagnosis of breast cancer.
Computing now stands at the threshold of another major leap forward building on the convergence of three separate developments: the continuing surge of computing power at affordable prices; the miniaturization of electronics, and the beginnings of the high bandwidth wireless digital communication. One consequence of the increased power and decreased size of computational devices is that autonomous robots with computers on board can process signals and images in real time, follow paths through a cluttered environment, avoid moving obstacles, and simultaneously interact effectively with humans in the environment. Applications of these capabilities are illustrated by the robot-guides of Burgard et al., of Mizoguchi et al., and of Nourbakhsh et al. Another type of application is exemplified by the fact that medical image sequences can be analysed in reasonable times, as in the XMRI software package of Hayton et al.
Leveraging the new power easily available in computing, AI can make the use of computers far more effective. Fromherz et al. describe how model-based computing, that has its origins in constraint programming, can provide the basis for a new software methodology that has supported flexible design, analysis and control of complex electro-mechanical equipment. Brachman et al. describe one practical implementation of the Classic knowledge representation system that has supported four billion dollars worth of purchases for AT&T. Chien et al. discuss experience with a system to plan and schedule shuttle payload operations, while Sparck-Jones reviews the interactions between information retrieval and AI. Muggleton describes results using inductive logic programming in molecular biology, and outlines the prospects for learning language. Finally, Forbus et al. review the development of CyclePad, a system that facilitates students learning about thermodynamic cycles, stressing the design and qualitative behavior of