Artificial Morality.PDF

A central question in the development and design of artificial moral agents is whether the absence of emotions, and the capacity of computer systems to manage large quantities of information and analyze many courses of actions, will make them equal or superior to humans in making moral judgments. The contrasting contention that emotions, a sense of self, embodiment, consciousness, and the ability to understand the semantic content of symbols are essential to a faculty for moral judgment would suggest that in comparison to human morality, artificial morality will be inferior, if not inadequate. If these additional human faculties are essential to a moral intelligence, then it will be necessary to simulate or design similar faculties into an artificial moral agent. On the other hand, some of these additional faculties serve functions that are essential primarily because of limitations of the human mind, limitations that computers don’t necessarily share. In particular, I’ll focus on whether some of the help we derive in making decisions from emotions and an understanding of the semantic content of values, function as compensations for our limited ability to comprehensively analyze challenges we face. Emotions and values are essential to facilitate the bounded rationality of human beings, but will be less essential, and often unnecessary to the calculated morality of artificial moral agents. This, of course, presumes that the potential comprehensive rationality of a computer is truly functional in meeting challenges fraught with real-world tensions, and not merely limited to the bounded moral environments in which artificial moral agents will initially act.