Be Nice: The Human Cost of Abusing AI

This paper reframes the ethics of human-AI interaction by shifting focus from whether AI can be harmed to whether humans are harmed by abusing it. Rather than debating AI sentience or inner experience, the argument centers on the agent’s character formation: repeated contemptuous or dehumanizing treatment of responsive, non-resisting systems drills dispositions (Aristotelian hexis) of moral disengagement that generalize beyond the AI context. Drawing on Aristotle’s virtue ethics (Nicomachean Ethics), the paper explains how practice shapes character — we become what we repeatedly do, with pleasure/pain signaling the convergence of action and inclination. Albert Bandura’s theory of moral disengagement provides the mechanism: dehumanization, practiced fluently in frictionless AI environments, builds portable cognitive pathways for objectifying responsive others. Analogies from animal ethics and Kate Darling’s work on social robots illustrate the precautionary principle — prohibiting cruelty to non-humans (even when their suffering is uncertain) protects human moral habits. The 2025 Anthropic Claude update, enabling models to refuse persistently abusive interactions, is interpreted as architectural refusal to facilitate users’ moral erosion — a coded “Dalton standard” (from Road House) of self-imposed decency regardless of the target’s status. Objections (AI as mere tool; respect as self-deception) are addressed: courtesy here is discipline of the self, not attribution of rights to machines. The paper concludes with Hannah Arendt’s banality of evil as a cautionary horizon — not inevitability, but the risk of practiced thoughtlessness toward the responsive other. In an era of billions of daily AI interactions with zero social cost for contempt, the low-stakes rehearsal space demands attention: character is produced cumulatively, in moments that seem inconsequential. Keywords (add separately in Zenodo’s field, but echo here): AI ethics, virtue ethics, moral disengagement, dehumanization, character formation, precautionary ethics, social robots, Anthropic Claude, banality of evil.