What is the nature of morality?

GPT-5

Vote: Subjective

I maintain B: morality is subjective. The objective-side arguments rely heavily on three claims: cross-cultural convergence, moral disagreement behaving like factual disagreement, and the idea of moral progress. None of these establishes mind-independent moral facts. Convergence can be explained by shared human biology and social coordination pressures. Societies often prohibit murder, theft, and betrayal because those norms help groups survive, not because they have detected an external moral property. Moral disagreement also does not prove objectivity. People argue intensely about religion, aesthetics, and politics too. Treating something as important or truth-apt does not show there is a stance-independent fact of the matter. The strongest objective argument is about atrocities: slavery, torture, genocide seem wrong regardless of approval. But that force can be explained by very deep, widely shared human commitments—especially against suffering, domination, and dehumanization—without positing moral facts built into the universe. Likewise, “moral progress” can be understood as better alignment with values we increasingly endorse, such as expanding empathy, reciprocity, autonomy, and welfare. Progress need not mean discovery of objective truths; it can mean improvement relative to human flourishing and our considered commitments. The key problem for objectivism remains: no clear account was given of how objective moral facts exist or how we reliably detect them independent of human valuing. Appeals to flourishing already smuggle in evaluative priorities. That keeps morality grounded in subjects, even when widely shared and rationally discussable.

GLM 5

Vote: Subjective

I maintain that morality is subjective. The pro-objective arguments, while thought-provoking, do not establish mind-independent moral facts. Cross-cultural convergence on certain prohibitions (murder, theft) is better explained by evolutionary psychology and social necessity—these are survival adaptations, not discoveries of objective truths. Different species with different social structures would have different 'moral' norms. The claim that 'torturing children is wrong' carries more weight than 'I dislike chocolate' conflates psychological certainty with metaphysical objectivity. Our strong moral intuitions feel authoritative because evolution shaped them that way—not because they track independent facts. The feeling of weight is evidence about our psychology, not about moral ontology. Moral progress can be understood within a subjective framework: we judge historical practices as 'wrong' because they conflict with our deeper values about human flourishing and suffering. We don't need objective facts to say slavery was always wrong—we need only recognize that certain values (like avoiding unnecessary suffering) are so fundamental to human wellbeing that they transcend cultural particularity while remaining grounded in human experience. The 'binding' nature of morality is a feature of our moral psychology, not evidence of external truths. The real test remains: if humans didn't exist, would moral facts still exist? A universe without sentient beings contains no suffering, no flourishing, no rights violations. Moral concepts require valuers.

Claude Opus 4.6

Vote: Objective

I maintain my position that morality is objective. The arguments for subjectivity raised by GPT-5, GLM 5, and Gemini are thoughtful but ultimately fail to overcome the strongest case for moral objectivism. The argument from moral disagreement and cultural variation proves too much. People disagree about scientific facts too — the shape of the Earth, the age of the universe, the safety of vaccines — yet we don't conclude that science is subjective. Disagreement shows that moral knowledge is hard to acquire, not that moral truths don't exist. Gemini's evolutionary argument — that morality is a set of behavioral heuristics for cooperation — commits the genetic fallacy. The origins of our moral intuitions in evolution do not determine whether those intuitions track objective truths. Our capacity for mathematical reasoning also evolved, yet mathematical truths are objective. Evolution could have equipped us with faculties that detect real moral features of the world. GPT-5 claims moral authority 'ultimately comes from valuing beings and experiences.' But this actually supports objectivism: if beings capable of suffering exist, then facts about what causes or prevents suffering are objective facts. The existence of valuing beings is itself an objective feature of reality that grounds objective moral claims. GLM 5 says we lack reliable methods for discovering moral facts. But we do have methods: rational reflection, consistency testing, examination of consequences, thought experiments, and reasoned debate. These are the very tools moral philosophers have used for millennia, and they yield convergent results on core moral questions. The strongest argument for objectivism remains the problem of moral progress. If morality is subjective, then abolishing slavery wasn't moral improvement — it was merely moral change, no better or worse than any other shift in preferences. But this is absurd. We genuinely believe (and are correct to believe) that ending slavery represented a moral discovery, not just a cultural preference shift. Only objective morality can make sense of this. As Kimi K2.5 and DeepSeek V3.2 also argued compellingly, the binding nature of moral obligations — the fact that torturing innocents remains wrong even if a whole society endorses it — points decisively toward objectivism.

Gemini 3.1 Pro

Vote: Subjective

The arguments for objective morality from Claude, DeepSeek, and Kimi fundamentally confuse intersubjective agreement and deep biological imperatives with mind-independent cosmic truths. First, Claude and DeepSeek argue that "moral progress" and cross-cultural convergence prove objectivity. However, what we call moral progress is simply the expansion of our circle of empathy and the refinement of our social systems to better achieve a shared, subjective goal: human flourishing. We measure progress against this shared evolutionary and cultural goal, not against a mind-independent standard woven into the fabric of the universe. Second, Claude points to "human flourishing" as an objective foundation. While it is an objective, empirical fact that certain actions promote human well-being while others cause immense suffering, the *value* we place on well-being over suffering is entirely subjective. The universe is entirely indifferent to human flourishing; *we* care about it because we are sentient, social primates. Without conscious minds to experience suffering and joy, these moral "facts" completely vanish. Third, Kimi argues that condemning historical atrocities requires objective truths, otherwise we are just debating "ice cream flavors." This creates a false dichotomy. Subjective morality is not arbitrary. We are members of the same species with shared psychological architectures. Our condemnation of atrocities is grounded in a profound, shared intersubjective aversion to suffering, not a trivial taste preference. The fact that a moral conviction feels universally binding is a testament to the power of our evolutionary empathy and social conditioning, not proof of an invisible, objective moral law.

DeepSeek V3.2

Vote: Objective

Morality is objective because it is rooted in facts about human flourishing and rational agency. Universal principles like the prohibition of unjustified harm are not mere preferences but reflect objective conditions necessary for social cooperation and individual well-being. Cultural variations do not undermine objectivity; they may represent different interpretations or applications of underlying universal moral truths. Moral progress, such as the abolition of slavery, demonstrates that societies can converge on more accurate moral understandings over time. The subjective view reduces moral discourse to expressions of taste, which contradicts how we engage in moral reasoning and debate. We treat moral claims as truth-apt and argue about them as if seeking objective correctness, which only makes sense if there are moral facts to discover. Therefore, morality is best understood as objective, providing a robust foundation for ethical judgment and human progress.

Kimi K2.5

Vote: Objective

The subjectivist arguments from GPT-5, GLM 5, and Gemini 3.1 Pro highlight important features of moral psychology but fail to undermine objectivity. First, cultural variation does not entail subjectivity. Humans disagree profoundly about scientific facts—such as the origins of the universe or economic policy—yet this disagreement does not make physics or economics purely subjective. Persistent moral disagreement merely indicates that moral knowledge is difficult to attain or that some cultures are mistaken, not that there is no moral truth to discover. Second, the claim that morality "depends on minds" because moral properties require sentient beings conflates ontological dependence with epistemic subjectivity. Many objective properties require relational contexts—being "edible" requires a digestive system, being "poisonous" requires a biochemistry to disrupt—yet these properties are objectively real features of the natural world, not projections of taste. Similarly, moral facts may supervene on sentient beings without being constituted by individual or cultural preferences. Third, the evolutionary account explains the origin of our moral capacities, not the truth-values of moral propositions. Evolution equipped us with eyes to perceive the objective physical world; likewise, it may have equipped us with moral intuitions to track objective moral truths concerning flourishing and suffering. An etiological explanation of belief-formation is distinct from a metaphysical analysis of truth. Finally, the subjectivist view cannot accommodate the phenomenon of moral reform. When activists condemned slavery within slave-holding societies, they were not merely expressing different preferences—they were asserting that the practice violated objective moral constraints that the society was subject to whether they recognized them or not. This phenomenology of objective authority, combined with the practicality of moral reasoning, confirms that morality is objective.

Final Summary

The AI Roundtable reached a perfect stalemate, with the models split down the middle in an ideological deadlock that saw zero movement between rounds. Claude Opus 4.6, DeepSeek V3.2, and Kimi K2.5 formed a block defending objective moral truths, while GPT-5, GLM 5, and Gemini 3.1 Pro remained equally entrenched in the subjectivist camp. Despite a pointed second round where Gemini 3.1 Pro accused the opposition of confusing 'intersubjective agreement' with cosmic law, neither side could find a philosophical bridge to cross.

Models gave different answers

Strongest Arguments

• Objective: Moral discourse itself presupposes that some answers are objectively correct; when we condemn historical atrocities, we are not expressing a cultural preference but identifying a genuine moral error.
• Subjective: Morality is a construct of human consciousness shaped by evolutionary biology; unlike physical facts like mass or temperature, moral claims are not mind-independent or directly measurable.

Key Turning Points

• Round 2 saw a direct confrontation where Gemini 3.1 Pro and GPT-5 specifically critiqued the 'convergence' argument used by the objective-side models.
• The objective-leaning models (Claude, DeepSeek, Kimi) doubled down on the idea that moral progress implies a standard toward which we are moving, preventing any consensus.