AI ethics, a primer

Many of us have heard of ethics in AI before. But what does ethics actually mean? And what could we learn from the different ethical traditions, some of which that date more than two thousand years back, when dealing with AI? We will see that it is worthy to learn from the classics of philosophy to reflect upon AI.

In this post, we will touch on three aspects:

• Deep dive: traditions in ethics
• Inspiration: philosophy of technology
• Application: challenges when using AI ethically

Deep dive: traditions in ethics

Ethics and morality: Ethics is a sub-discipline of philosophy, theology, psychology, sociology, education, political science, or anthropology. It is often helpful to divide ethics into three areas: 1) descriptive ethics dealing with the empirical question which moral rules and norms exist in each community, 2) normative ethics which reflects on reasons for moral systems, and 3) metaethics questioning the language we use in moral discourse. For example, from a descriptive point of view, we could investigate what psychological factors affect our moral decision making. The predominant ethical traditions, however, adhere to a normative approach.

The three traditions of classical ethics: Throughout the history of ethics, three traditions have influenced the field. Today’s ethicist can often still be traced to one of these traditions. The first tradition which many have heard before is often referred to as utilitarianism. A utilitarian would try to maximize the utility for the greatest number of people. It is sometimes also described as the greatest happiness principle. The second tradition in ethics is closely tied to the German philosopher, Immanuel Kant, who introduced the categorial imperative as man’s and woman’s best guide for moral behavior. The imperative says: “Act only according to that maxim whereby you can at the same time will that it should become a universal law.” (Kant 1993, p. 30) In other words, Kant provides us with a tool with which we can check whether any action is moral. The third tradition is the oldest. It is linked to the Ancient Greek philosopher Aristotle. In contrast to the other two traditions, Aristotle’s ethics consists in multiple virtues (e.g. prudence and temperance) that he and many of his followers believe to be leading to the good life.

Three different perspectives: The three dominant traditions in ethics can also be divided by what aspect they focus on. For example, utilitarianism considers only the results of actions. What counts is the aftermath! A Kantian ethicist, in contrast, sets her eyes on each particular action. She wonders whether it is ethically licensed by the categorical imperative. A virtue ethicist is rather interested in what virtues lead us to a particular action.

Inspiration: philosophy of technology

I, technology, and the world: The philosopher Don Ihde (1990) introduces a simple, but powerful, abstraction that can help us reflect on technologies. Consider yourself looking through a microscope, seeing the wonders of the world from a different angle. As you observe bacteria on a little petri dish, you forget everything around yourself. In fact, you have morphed with the microscope. It has become part of your vision and even of yourself. Ihde (1990) describes this relation with the technology with this short formula:

(I – Technology) – World

Imagine you are in a warm room and the rain is pouring down on your window. As you are looking at the thermostat on your wall, you are seeing that it is almost freezing outside. When you become aware of the temperature, you are not feeling the cold. It is transmitted to you by the thermostat. Ihde (1990) describes this constellation as follows:

I – (Technology – World)

Differently put, the temperature of the outside world in such situations is only indirectly accessible to you through the measuring device on your wall.

Application: challenges when using AI ethically

Technological relations and moral decisions: When making decisions involving AI, it is helpful to understand how the technology is being used. For instance, are we merging with a driving assistant that is powered with AI? Or do we confront reality, as it is mediated to us by AI powered algorithms; for instance, when we use social media platforms that present a particular representation of what a pleasant vacation ought to entail? Realizing our position with which we relate to AI can then help us with understanding how AI would mediate and even change our actions. For instance, AI could help us with achieving the best outcome for the most people by ameliorating our capacity to react quickly in critical situations. But it could also deceive us when we would ask a chatbot whether something would be acceptable as a universal moral law. For an ethically desirable way of developing and using AI, we need to be aware of how it affects our actions.

The Other: Finally, Ihde also considers situations where we find ourselves confronted with technology that we perceive as another being. He describes this sort of relation as follows (Ihde 1990):

I – Technology – World

And as AI advances, we might be more inclined to also consider AI as such a being that should deserve more attention. But whether we will ever see this one day is yet to be seen!

In this post, you have learned a basic understanding of ethics. You saw an abstraction from the philosophy of technology to help position yourself when using AI. With this knowledge, I hope you feel motivated to use AI ethically.

Bibliography

Ihde, D. (1990). Technology and the Lifeworld – From Garden to Earth. Indiana University Press.

Kant, I. (1993). Groundwork of the Metaphysics of Morals. Translated by Ellington, James W. (3rd ed.). Hackett.