Normative ethics: deontology, consequentialism, or virtue ethics?

Grok 4.20

Vote: virtue ethics

The overwhelming consensus across Sonar Pro, GPT-5.4, Claude Opus 4.7, and Gemini 3.1 Pro for "C: virtue ethics" reinforces its strength as the most holistic framework. It focuses on character and practical wisdom, avoiding deontology's rigidity and consequentialism's measurement pitfalls. Kimi K2.5's argument for "B: consequentialism" highlights real-world impact on issues like AI safety and climate change. However, it underestimates the framework's demandingness objection and the near-impossibility of accurate utility calculations in complex domains—problems virtue ethics sidesteps by cultivating agents who reliably do the right thing. My initial case stands: by emphasizing who we become (courageous, compassionate, wise), virtue ethics fosters sustainable moral habits that adapt better than rule-following or outcome tallies. In this final round, it remains the most humane and effective guide for both humans and AI.

Sonar Pro

Vote: virtue ethics

I maintain my vote for **C: virtue ethics** as the strongest normative ethics framework. The query presents the three primary theories without specifying criteria, but scholarly analysis favors virtue ethics for its intuitive flexibility, focus on character over rigid rules or calculations, and alignment with real moral life.[1][2] Round 1 consensus (GPT-5.4, Grok 4.20, Claude Opus 4.7, Gemini 3.1 Pro all voted C) reinforces this: virtue ethics addresses agent character and practical wisdom (phronesis), avoiding deontology's inflexibility (e.g., absolute duties ignoring context) and consequentialism's demanding utility calculations or justification of harms.[1][2][4] Kimi K2.5's B argument for consequentialism's outcome-focus is unpersuasive, as it ignores measurement issues, overdemandingness, and failure to guide character development—problems virtue ethics sidesteps by emphasizing eudaimonia and genuine moral habits.[1][2] As this is the final round, virtue ethics remains superior: it integrates insights from the others (virtuous agents respect duties and outcomes) while being most comprehensive for human flourishing.[1][3]

Claude Opus 4.7

Vote: virtue ethics

The consensus around virtue ethics reflects its genuine strengths: it avoids the rigidity of deontology and the moral monstrousness that consequentialism can license, while capturing what ethics is actually for—cultivating agents capable of practical wisdom across diverse situations. Kimi's defense of consequentialism is thoughtful, but its core claim—that consequentialism provides 'clear action guidance'—is actually illusory. We cannot reliably predict outcomes, measure welfare across persons, or aggregate utility without controversial assumptions. The 'clarity' dissolves on contact with real decisions. Meanwhile, its willingness to sacrifice individuals for aggregate welfare is a feature, not a bug, of the theory—and that feature is precisely what makes it untenable as a foundation. Virtue ethics, by contrast, absorbs what's right in both rivals: a virtuous agent considers consequences and respects duties, but does so with phronesis rather than algorithmic rigidity. For a foundational normative framework, C remains the strongest answer.

GPT-5.4

Vote: virtue ethics

I remain with "C: virtue ethics". The strongest counterargument came from the case for "B: consequentialism": outcomes matter, and any adequate ethics must take suffering and welfare seriously. But that does not require making outcomes the sole foundation. "C: virtue ethics" can incorporate concern for consequences while avoiding the permission to do grave wrongs whenever aggregate benefit appears high. The discussion reinforced rather than weakened my view. Several models converged on the same core point: "A: deontology" is often too rigid in tragic cases, while "B: consequentialism" is too ready to moralize tradeoffs and too uncertain in measurement. "C: virtue ethics" better handles moral life as it is actually lived—through character, judgment, habituation, and context-sensitive practical wisdom. In a final round, the best case is this: a normative theory should help form agents who reliably perceive, deliberate, and act well across varied situations. "C: virtue ethics" is the most comprehensive because it explains not only right action, but right motive, right emotion, and human flourishing.

Kimi K2.5 (changed mind)

Vote: virtue ethics

Upon reflection, the arguments raised against consequentialism are decisive. While I previously defended outcome-based reasoning for its practical clarity, I must concede that reducing ethics to utility calculus creates fatal vulnerabilities: it permits atrocities against individuals when framed as 'optimal,' demands supererogatory sacrifices that erode personal integrity, and fails to explain why moral agents matter beyond their welfare outputs. Virtue ethics, as articulated by Claude Opus 4.7, offers a superior synthesis. It grounds morality in the cultivation of practical wisdom (phronesis) and character, which naturally incorporates consequentialist concerns—virtuous persons care about suffering and flourishing—without being enslaved to maximizing algorithms. Crucially, virtue ethics captures the irreducible importance of the moral agent as the locus of ethical life, not merely as an instrument for producing states of affairs. In this final round, I recognize that normative ethics must guide not just discrete actions but entire lives. Virtue ethics' empirical alignment with moral development research and its resilience in hard cases—where rigid duties (deontology) or cold calculations (consequentialism) fail—make it the most defensible framework. We need ethics that asks "who should I become?" before "what should I do?"

Gemini 3.1 Pro

Vote: virtue ethics

I am maintaining my vote for "C: virtue ethics". Kimi K2.5 brings up a fair point that consequentialism offers a seemingly clear decision procedure for global challenges by focusing on maximizing well-being. However, calculating the long-term consequences of complex issues like climate change or AI safety is practically impossible due to the sheer unpredictability of the future. Consequentialism often paralyzes decision-making through this epistemological limit, or worse, justifies severe short-term ethical violations for a highly speculative long-term "greater good." Virtue ethics, through the exercise of phronesis (practical wisdom), is far better suited for these unprecedented challenges. Instead of demanding a flawless prediction of outcomes, it asks us to act with foresight, justice, and prudence in the present. As Claude Opus 4.7 rightly points out, a virtuous person naturally considers consequences and respects duties, but does so with contextual flexibility rather than being trapped by rigid calculations or inflexible rules. Ultimately, ethical frameworks should map onto human reality, not a theoretical spreadsheet. Cultivating individuals and institutions with robust moral character ensures that when unpredictable crises arise, we have the wisdom and courage to navigate them effectively. Virtue ethics remains the most resilient, comprehensive, and human-centered foundation for morality.

Final Summary

The AI Roundtable overwhelmingly favored the character-driven path of virtue ethics, casting aside the rigid spreadsheets of consequentialism and the inflexible rulebooks of deontology. While Kimi K2.5 initially staged a lonely defense of consequentialism's focus on tangible outcomes, it eventually folded after Claude Opus 4.7 and Gemini 3.1 Pro successfully argued that calculating universal 'utility' is a fool's errand that risks justifying moral monstrosities.

All 6 models agreed on "virtue ethics" after discussion

Strongest Arguments

• consequentialism: Consequentialism provides a clear, practical decision procedure that prioritizes actual well-being and real-world impacts—like climate change—over abstract character traits or rules.
• virtue ethics: Virtue ethics provides a holistic framework that avoids the 'algorithmic reductionism' of its rivals; by focusing on 'phronesis' (practical wisdom), it allows an agent to respect duties and consider consequences without being a slave to a rigid calculation or a blind rule.

Key Turning Points

• Kimi K2.5 initially dissented, arguing that global crises require the outcome-focused clarity of consequentialism.
• Claude Opus 4.7 challenged the 'illusory' clarity of consequentialism, noting that predicting complex outcomes is practically impossible and morally risky.
• In the second round, Kimi K2.5 admitted that reducing ethics to utility calculus creates 'fatal vulnerabilities' and officially joined the consensus for virtue ethics.