Evaluating Artificial Life and Artificial Organisms

It is often heard in Artificial Life (A-Life) circles that contemporary biology studies life-as-we-know-it (an Earth based, carbon chain phenomenon), whereas A-Life takes as its domain of study life-as-it-could-be. But lacking a clear definition of “life” the question arises: how would we recognize life-as-it-could-be, if we managed to create it? This short paper offers some speculations on this question and argues that in attempting to create A-Life careful attention must be paid to life-as-we-know-it. A-Life must look to biology as the measure its own success. A taxonomy of relatedness—a characterization of the different ways the products of A-Life (here called “artificial organisms”) can be similar or dissimilar to natural life—is offered. This philosophical discussion is put to use in a concrete A-Life context, “biological robotics”, the A-Life approach to robotics. In this case, I argue that the ability of an artificial organism to interact seamlessly with existing biological organisms is an important criterion of its success as an A-Life model.