On Vallor, "Technology and the Virtues"; Wong and Cua on Confucian ethics.
The question. If AI governance requires not just good rules but good institutional character, what does that imply about how regulatory bodies, companies and civil society organisations should be designed and staffed?
Rules and character
Virtue ethics asks what kind of people and institutions we should be. Morals are the external codes — the specific rules of right and wrong, learned from a society or a religion. Virtues are the internal traits that make someone follow them.
The three classes line up as three questions. Long-termism: what produces the best outcomes? Virtue ethics: what kind of people should we be? Kant: what do we owe people as persons?
Vallor argues that in all three classical traditions, rigid rule-following and utility-maximising are at best subservient and conditional aspects of living well, and at worst mere semblances of virtue. The virtue does not serve the rule; the rule is a rough guide a cultivated person can override when the situation calls for it. Cua's version of the same point: Confucian ethics has no rules of relevance and no rules of inference, and the junzi is paradigmatic because he copes with changing circumstances rather than applying a procedure.
The bow, and the institution
Vallor uses the bow and arrow — which made killing at a distance possible — to show that a technology changes not only the tactical circumstances but the moral landscape. If virtues are formed by repeated practice, and a technology decides which practices are cheap and easy to adopt, then emerging technologies are shaping our characters whether or not anyone intended it.
That may hold for a person. Institutionally it meets harder weather. The competitive dynamics of the AI race push frontier labs to underfund alignment, because the cost of falling behind is unimaginable. Star researchers leaving OpenAI, Google reversing on Project Maven — both are a failure of moral responsibility at the level of the institution. Anthropic was founded on a disagreement about safety, and both companies now compete for the largest IPO in history.
Do those founding virtues hold against the pressure to raise more? Or must the race to AGI be won at all costs, including the occasional black ball out of the urn? It looks as though the personal characters of a few lab leaders are being projected onto a society, whatever individual heroic objections are raised along the way.