The Ethical Computer Grows Up

AUTHOR

Donald Gotterbarn (USA)

ABSTRACT

Terry Winograd [Winograd 1991] argued that ‘ethics’ and ‘values’ are not the kinds of things addressed with computer science theories, but they are “…a domain in which we interpret and address our actions as professionals”. Rather then talk about observing ethics he talks from the point of view of the participant in ethics and describes three models of doing computer ethics. He characterized one model as “the angel or devil debate”, a model where you know what is right but have to exercise the moral fiber to resist the temptation to do what is wrong. The primary reason that this is an inadequate model for computer ethics is that many of the issues of computer ethics are not clear but they have to be worked out before we could exercise the simple decision between the devil and the angel. A second model he described is one that might help with the analysis, it the “Morality Computer”. In this model, the moral rules of computer ethics are part of the internal program logic of a computer. One does ethics by entering into the computer data describing a moral situation and the computer applies the rules to the situations and the computer cranks out a decision that is a result of applying the ethical axioms to the moral situation. The Computer helps you apply the known rules in this case. This model, however, cannot adjust to the situations for applying these rules. The model does not take into account the dynamics and social interaction of the ethical situation.

This ethical computer approach is reminiscent of the approaches sometimes taken in philosophy. One could talk of the philosophical computer in which the moral rules were the principles of some philosophical theory. A utilitarian theory would for example use the computer to calculate the ethical thing to do by doing a hedonistic calculus. Philosophers and others have recognized this as a mistake. Nevertheless, computer scientists quite readily take up this philosophical computer concept. The author of this piece has co-authored an article [Anderson 1993 “Using the ACM Code of Ethics in Decision Making”] which loaded the computer with the ACM Code of Ethics and used it to show how ethical decisions could be made by applying the Code of Ethics.

Winograd’s choice for a model of computer ethics is based on his emphasis that ethics is an activity involving social interaction that changes as each new stakeholder is involved and that the actions of one stakeholder will change the situation for all of the other stakeholders. He calls the process of ethical decision making in this environment ‘Ethicing’ to emphasize that ethics is primarily an activity. He models ‘Ethicing’ with a troop of jugglers. Not only are they involved in constant activity but also they must incorporate evolving standards and practices.

Winograd is right about the social dynamic of the way in which we do ethicing, but I think he was mistaken to totally abandon the notion of the morality computer. He his right that the model of actually having the computer make the ethical decisions is incorrect, but the Morality Computer can still function as a tool to help us ensure that we do not overlook anything in the decision making process. The juggling model of ethical decision makers has the weakness of including significant distractions. The ethical deliberations may be incomplete or fail to account for significant stakeholders. We explore the hypothesis that a morality computer, which provides the form for ethical decision-making, can be useful in computer ethics.

The decision procedures examined and implemented include the basic decision elements of ethicing identified by McFarland [1991]. The minimal elements include gathering of facts, analyzing the data in a reasonable and logically consistent fashion, and basing the decision on sound ethical principles. The inclusion of all of this in a particular judgment is what is needed. Going through this process manually is a tedious and subject to error.

There are numerous ethical decision making processes. For a collection of process used since see Maner [1999]. Recently, two processes have been focused upon because of the way in which they meet McFarland’s criteria. These methods are Collin’s Paramedic method and Kvanvig’s Resolvedd method

In work done at the Software Engineering Ethics Research Institute we have developed software tools to automate these process. This should not be confused with Winograd’s “Morality Computer “. The software tool developed facilitate the correct use of these processes in a synergistic situation which might lead the ethical decision maker to ignore critical steps in the decision making process. These tools have been applied to several ethical cases. The paper will consist of demonstrating these tools and evaluating the underlying ethical decision methodology. The tools application to two cases will be demonstrated and the strengths and weakness of these models as revealed by the automation of the decision and its application to these two cases. It is believed this will show that a morality computer, which provides the form for ethical decision-making, can be useful in computer ethics. The software will be made available for use and research by other scholars.

REFERENCES

Anderson, Johnson, Gotterbarn, and Perrolle, “Using the New ACM Code of Ethics in Decision Making,” Communications of the ACM, 1993

Collins, Miller, “The Paramedic Ethics for Computer Professionals,”

Kvanvig, J.L., “The Resolvedd method,” 1999

Maner, “Heuristic Methods in Computer Science Decision,”1999
McFarland, “Urgency of ethical standards intensifies in computer community,” IEEE Computer 1990.

Winograd, “Computer, ethics, and Social Responsibility,” Computing and Human Values: Proceedings of the 1991 Conference, Eds. Terry Bynum and Walter Maner.