The Scientific and Philosophical Scope of Artificial Life

Abstract

The new interdisciplinary science of ALife has had a connection with the arts from its inception. This paper provides an overview of ALife, reviews its key scientific challenges and discusses its philosophical implications. It ends with a few words about the implications of ALife for the arts.

Artificial life (ALife) is a young interdisciplinary collection of research activities aimed at understanding the fundamental behavior of lifelike systems by synthesizing that behavior in artificial systems. As befits a journal for artists who use science and developing technologies, Leonardo regularly publishes articles discussing ALife. There is also traffic in the other direction: for example, the biennial International Conference on ALife is the primary vehicle for publishing the latest scientific developments in ALife. Nevertheless, more than 5% of the articles published in the proceedings of the most recent ALife conference [1] concerned the application of ALife to art and music [2-5]. People in both communities believe that the arts and ALife have much to offer each other. Given this mutual interest, it would useful for the two communities to come to know each other better. The resulting opportunity to counteract the hype and misleading publicity surrounding ALife would also be welcome. The truth about ALife is often more interesting and surprising than the fiction and is always more valuable.

This paper aims primarily to provide an overview of ALife, explaining its approach to science and technology, outlining its major open problems and sketching its broader philosophical implications. It ends with a few words about the implications of ALife for the arts.

An Overview of Alife

Life is an interconnected web of adaptive systems produced spontaneously by the process of evolution. Living systems exhibit impressively robust and flexible functionality at many levels of analysis. Examples range from the genomic and proteomic [6] regulatory systems that control how biological organisms develop and function to the evolving ecological networks through which members of different species interact. Human-made adaptive systems, such as the myriad communication networks that span the globe, are beginning to approach the complexity of the adaptive systems found in nature. Learning how to engineer flexible and robust adaptive complexity is one of the biggest challenges facing society in the 21st century.

Traditionally, adaptive systems of various kinds were modeled independently in different disciplines. ALife is now bringing together biologists, physicists, chemists, psychologists, economists and anthropologists with computer scientists, philosophers and artists to create a unified understanding of adaptive systems of all types. ALife studies life and lifelike processes by synthesizing them in artificial media, most often using computer technology. The goals of this activity include modeling and even creating life and lifelike systems; the goals also include developing practical applications involving new technologies that exploit intuitions and methods taken from living systems. The phrase "artificial life" was coined by Christopher Langton. He envisioned a study of life as it could be in any possible setting and organized the first conference to explicitly address this field [7]. There has since been a regular series of conferences on ALife and a number of academic journals have been launched to publish work in this new field.

ALife borrows from other, older disciplines, especially computer science, cybernetics, biology and the study of complex systems in physics. Its closest intellectual cousin is artificial intelligence(AI). There is, however, a crucial difference between the modeling strategies that AI and ALife typically employ. Most traditional AI models are top-down-specified serial systems involving a complicated, centralized controller that makes decisions based on access to all aspects of global state. The controller's decisions have the potential to affect directly any aspect of the whole system. On the other hand, many natural living systems exhibiting complex autonomous behavior are parallel, distributed networks of relatively simple low-level "agents" that simultaneously interact with each other. Each agent's decisions are based only on information about its own local situation, and its decisions directly affect only that situation. ALife's models follow nature's example. The models themselves are bottom-up-specified parallel systems of simple agents interacting locally. The local interactions are repeatedly iterated and the resulting global...