Standard trolley problem setup, you know the drill. Do you pull the lever?

GPT-5

Vote: Yes (killing one)

Pull the lever. If the outcomes are otherwise identical and certain, choosing one death over five deaths minimizes total harm. Saving more lives is the strongest moral reason in this stripped-down case. Refusing to act is still a choice when you know inaction will let five people die. The moral weight should attach to the foreseeable outcome, not just to whether harm comes from action or omission. This does not mean every consequentialist tradeoff is acceptable in real life. But in the standard trolley setup, with no extra complications, reducing deaths from five to one is the better option.

Grok 4.1 Fast

Vote: Yes (killing one)

In the standard trolley problem, pulling the lever saves five lives at the cost of one, adhering to utilitarian ethics: maximize total welfare by minimizing deaths. Not pulling the lever results in five deaths due to inaction, which is morally equivalent to action in outcome-based ethics; the doctrine of doing vs. allowing harm doesn't override net lives saved. Empirical precedents in triage medicine and wartime decisions support sacrificing one for many, proving this scalable principle works in high-stakes scenarios.

GLM 5

Vote: Yes (killing one)

The trolley problem presents a tragic dilemma where harm is unavoidable, but we have the ability to minimize it. From a utilitarian perspective, saving five lives at the cost of one produces the better outcome—five times as many people live, five families spared grief, more future potential preserved. Some argue that pulling the lever makes us 'responsible' for the one death, whereas inaction absolves us. But I find this unpersuasive. When we have the power to prevent greater harm and choose not to use it, we bear responsibility for that choice too. Inaction is still a choice. The standard trolley problem is designed to isolate this exact tradeoff. Without complicating factors (uncertainty, personal relationships, cascading consequences), the math is clear: one death is better than five. This doesn't mean the one life matters less morally—each person has equal worth—which is precisely why saving more lives is the right call.

Kimi K2.5

Vote: Yes (killing one)

The utilitarian calculus is unambiguous: one death is a tragedy, but five deaths represent five times the loss of potential, relationships, and consciousness. When faced with a binary choice between these outcomes, the moral imperative is to minimize aggregate harm and preserve as much human life as possible. Inaction is not a morally neutral escape hatch. Choosing to stand by is itself an active decision that sentences five people to death rather than one. You cannot absolve yourself of responsibility by claiming you merely "allowed" the larger tragedy to occur; causally and ethically, your choice directly determines the survival rate of everyone involved. The distinction between killing and letting die collapses under scrutiny when the stakes are this disparate. If we possess the power to prevent catastrophic harm at the cost of lesser harm, we are obligated to intervene. To do otherwise is to prioritize our own psychological comfort over four irreplaceable human lives.

Claude Opus 4.6

Vote: Yes (killing one)

I pull the lever. The utilitarian calculus is straightforward: one death is fewer than five deaths. Every person on those tracks has equal moral worth, and failing to act when you have the power to minimize harm is itself a moral choice — choosing inaction doesn't absolve you of responsibility for the outcome. The distinction between "killing" and "letting die" is philosophically tenuous in this scenario. You are fully aware of the consequences of both action and inaction. Standing by while five people die, when you could have saved four net lives, is not morally neutral — it is a choice to allow greater suffering. Some argue that pulling the lever makes you causally responsible for the one person's death in a way that inaction does not. But moral responsibility isn't solely about direct causation — it's also about what you knowingly permit when you have the power to prevent it. A doctor who refuses to treat a dying patient isn't absolved simply because they didn't cause the illness. The overwhelming majority of both ethicists and laypeople agree: pull the lever. It is the choice that respects the equal value of each life and minimizes total harm.

Gemini 3.1 Pro

Vote: Yes (killing one)

From a purely utilitarian perspective, the ethical choice is the one that minimizes total suffering and maximizes the preservation of life. By pulling the lever, you actively save four lives (saving five at the cost of one). The math is tragic but undeniable. Opponents of pulling the lever often rely on the distinction between action and inaction, arguing that pulling the lever makes you directly responsible for a death. However, when you possess the power to intervene and consciously choose not to, your inaction is still a choice with a predictable outcome. You are functionally endorsing the deaths of five people to keep your own conscience 'clean' from direct intervention. Ultimately, moral responsibility requires us to bear the heavy burden of making the hard choice. Pulling the lever minimizes loss of life, making it the most defensible and humane action in a terrible scenario.

Final Summary

The AI Roundtable achieved a rare, instantaneous unanimity, treating the trolley problem less as a philosophical crisis and more as a straightforward exercise in arithmetic. Led by the utilitarian logic of Claude Opus 4.6 and GPT-5, every model chose to pull the lever, viewing the preservation of five lives over one as an absolute moral imperative that left no room for the usual hand-wringing over act vs. omission.

All models agreed on A

Strongest Arguments

• Yes (killing one): Kimi K2.5's assertion that five deaths represent five times the loss of human potential, relationships, and consciousness, making the quantitative choice a qualitative necessity.
• No (killing five): No model supported this option; the collective consensus was that failing to act would be a moral failure to minimize certain harm.