Angels and artifacts: Moral agents in the age of computers and networks

Abstract

Traditionally, philosophers have ascribed moral agency almost exclusively to humans (Eshleman, 2004). Early writing about moral agency can be traced to Aristotle (Louden, 1989) and Aquinas (1997). In addition to human moral agents, Aristotle discussed the possibility of moral agency of the Greek gods and Aquinas discussed the possibility of moral agency of angels. In the case of angels, a difficulty in ascribing moral agency was that it was suspected that angels did not have enough independence from God to ascribe to the angels genuine moral choices. Recently, new candidates have been suggested for non‐human moral agency. Floridi and Sanders (2004) suggest that artificially intelligence (AI) programs that meet certain criteria may attain the status of moral agents; they suggest a redefinition of moral agency to clarify the relationship between artificial and human agents. Other philosophers, as well as scholars in Science and Technology Studies, are studying the possibility that artifacts that are not designed to mimic human intelligence still embody a kind of moral agency. For example, there has been a lively discussion about the moral intent and the consequential effects of speed bumps (Latour, 1994; Keulartz et al., 2004). The connections and distributed intelligence of a network is another candidate being considered for moral agency (Allen, Varner & Zinser, 2000). These philosophical arguments may have practical consequences for software developers, and for the people affected by computing. In this paper, we will examine ideas about artificial moral agency from the perspective of a software developer.