The Paradox of Virtue
The paradox of virtue refers to a problem in ethics where the practice of virtue by individuals can potentially lead to the destruction of that very virtue. Examples of the paradox of virtue include the following:
First, there is the problem that practicing virtue can potentially destroy it. For example, suppose a person possesses the 'virtue of valuing honesty.' If that person makes an honest confession in a certain situation, that confession might damage the trust or friendship of others. Therefore, by confessing honestly, they end up destroying the original virtue.
There is also the problem that practicing one virtue might potentially require sacrificing another. For example, suppose a person possesses the 'virtue of valuing fairness.' If that person ends up hurting others in order to exercise justice in a certain situation, that act becomes contradictory to fairness.
The issues surrounding the paradox and dilemmas of virtue are significant topics addressed by a wide range of scholars, from ancient Greek philosophers to modern ethicists. Contemporary virtue ethicists such as Alasdair MacIntyre, Martha Nussbaum, and Rosalind Hursthouse have engaged with problems concerning virtue dilemmas and paradoxes.
There are many factors beyond the paradox of virtue that make defining virtue and good or evil difficult in AI ethics. For instance, definitions and concepts of virtue vary greatly depending on historical background and religious influence. Furthermore, there are many scholars who, from positions like nihilism or skepticism, argue that there is no point in discussing virtue in the first place.
The author has organized the points of discussion into the following four main categories. Please note that the ethical theories listed below are not meant to represent individually established academic fields, but are broad categorizations used by the author for the purpose of organizing the issues, and should not be considered formal classifications in ethics.
(1) Machine Virtue Ethics refers to theories or doctrines that argue that AI itself requires virtue. Key proponents include Wendell Wallach, Shannon Vallor, and Colin Allen.
(2) Virtue Ethics for AI Developers refers to theories or doctrines that argue that AI developers require virtue. Key proponents include Shannon Vallor, Michael Anderson, and Susan Leigh Anderson.
(3) Combined Virtue Ethics for AI and AI Developers refers to theories or doctrines that argue that both AI and AI developers require virtue. A key proponent is Antonio Casilli.
(4) Moral Nihilism or Moral Skepticism refers to theories or doctrines that argue that discussing virtue is meaningless because it cannot be defined. While all skeptics fall under this category, skeptics often fall into the 'paradox of skepticism' described next, making it difficult to clearly identify who exactly is a proponent of this doctrine.
Discussions of virtue in AI ethics involve a diverse range of positions and theories that are intricately intertwined. Because there are approaches from different standpoints—such as historical backgrounds, religious influences, nihilism, and skepticism—in addition to the problem of the paradox of virtue, the difficulty of defining virtue and good or evil is evident. To grasp this complex situation and discuss AI ethics appropriately, it is important to maintain diverse perspectives.