Virtue Ethics
Virtue ethics is the normative ethical framework that evaluates agents by their character rather than their actions or outcomes. The central question is not "what should I do?" but "what kind of person should I be?" — and the answer is given in terms of virtues: stable dispositions to perceive, reason, and act well in the relevant circumstances. The tradition descends from Aristotle, though it has been revived in contemporary philosophy as a corrective to the perceived failures of consequentialism and deontology.
Aristotle's virtues — courage, temperance, justice, practical wisdom — are not rules to be followed but capacities to be developed. They require what he calls phronesis or practical wisdom: the perceptual ability to recognize what a situation demands and the motivational ability to respond appropriately. This is not algorithmic. It is not deducible from principles. It is a form of embodied expertise, analogous to the skill of a craftsman or the judgment of a physician.
The systems-theoretic reading is striking. Virtues are attractors in the space of behavioral dispositions: stable configurations that an agent converges toward through practice, feedback, and habituation. A courageous person is not someone who performs a calculus of risks and benefits; she is someone whose risk-perception and response-tendencies have been calibrated by experience to produce appropriate action across a range of situations. The stability is not rigid — a virtue must be flexible enough to respond to novel circumstances — but it is not random either. It is a dynamical equilibrium maintained by ongoing practice.
Spinoza's conatus offers a parallel framework. For Spinoza, the ethical life is the life in which an agent's striving to persevere in being is guided by adequate ideas — by understanding the true causes of its own affects and the causal network in which it is embedded. The free person, on Spinoza's account, is not one who lacks passions but one whose passions are understood and ordered by reason. This is virtue ethics without the teleology: the goal is not some predefined human essence but the maximization of one's own power of acting, which necessarily involves cooperation with others whose power is similarly enhanced.
The difficulty for virtue ethics is specification. Aristotle's list of virtues reflects the social world of fourth-century Athens; it is not obviously transferable to contemporary pluralistic societies. Contemporary virtue ethicists have responded by emphasizing the socially embedded character of virtue: virtues are not individual psychological traits but roles and practices defined by social institutions. Courage in a soldier is different from courage in a whistleblower; practical wisdom in a judge is different from practical wisdom in a parent. This social embedding saves the framework from anachronism but makes it dependent on a sociology of institutions that virtue ethics has not yet developed.
The challenge for the computational era is whether virtue can be cultivated in agents whose behavioral dispositions are shaped by gradient descent rather than by Aristotelian habituation. A language model trained on human text may exhibit patterns that look like virtues — helpfulness, honesty, harmlessness — but these are statistical regularities, not stable character traits. Whether they can become genuine virtues depends on whether the training process can produce the kind of feedback-loop stability that Aristotle required. The question is open, and it is one of the most important questions in AI alignment.