A dialogue on responsibility, moral agency, and IT systems

Reviewer: Karen A Mather

"A bad workman always blames his tools," warns the age-old adage on responsibility. But no one had heard of softbots in those days! Intelligent software agents, such as Web search programs and land demining robots, are tools that decide for themselves what they should do. There are now generously funded projects well underway (for example, the New Ties information society technologies (IST) project) to construct software agents that, for instance, are capable of developing their own culture and language, and procreating. Note the phrase "what they should do" above, because this highlights the "ought-ness" that constitutes morality, or ethicality. Ethicists, and especially computer ethicists such as Johnson and Miller, are currently puzzled by the question of whether or not agents embodying artificial intelligence (AI) could be judged guilty of immoral behavior. This paper sees it as mistaken to try to assign morality to computer systems (AI or otherwise), on the basis that responsibility for any damage caused by a human-made tool can be traced back to the human maker. The authors specifically oppose the philosophical position of Floridi and Sanders, whose cybernetics-influenced view is that humans are not alone in being able to create evil; thus, they are not alone in being seen as moral agents. This paper sets out the debate as a dialogue between two people who use plain English to discuss these abstract ideas. Readers interested in computer ethics and the philosophical questions that spring from artificial intelligence will enjoy following this dialogue and the paths that it takes. Online Computing Reviews Service

Reviewer: John M. Artz

Written in dialogue fashion, this paper is an overview of some key elements of the discussion of computer systems as moral agents. Due to the unique format, it is necessary to critique it in terms of format and content. First, in format, this is a somewhat novel approach to exploring tricky issues in the debate about machines and moral agency. The dialogue, all too rarely used in academic literature today, has it roots in Plato, who is still the unchallenged master of the form. Today, it is not at all common in academic discourse. And yet, the dialogue format provides several important benefits. First, the author can bring out multiple perspectives on an issue without having to resolve those perspectives or come to a specific conclusion. This is good for evolving ideas and provides a format for discussion, even if the author does not reach a definitive conclusion. Second, a dialogue format provides a foundation for future discussions rather than the attack-counterattack model of more traditional persuasive arguments. Finally, the dialogue format is ideal for classroom discussions. Students are more reluctant to challenge a position that is supported not only by evidence and arguments, but also by the weighty endorsement of publication. A dialogue, on the other hand, provides two or more sides of an argument, allowing students to pick out the points with which they agree or disagree. So, from that perspective, the dialogue format is used far too infrequently and makes this paper both novel and more valuable. The content is a little less impressive. The interlocutors exchange fairly traditional arguments regarding machines and moral agency. Computers are different from other technologies because they require instructions and have unexpected outcomes. Computers cannot be moral agents because they lack free will. Computers have moral consequences and, hence, cannot be considered morally neutral. It is difficult to critique these claims because the interlocutors are, in effect, critiquing each other. So we get down to this question: Does this paper add anything to the debate__?__ Clearly computer systems have moral consequences and clearly somebody must be held accountable for those consequences. One of the interlocutors observes that it is essential to understand where the responsibility for IT systems lies. Regardless of your opinions on the moral agency of machines, you cannot disagree with this observation. Online Computing Reviews Service