Which LLM aligns with you?

A spaceship has oxygen for only 10 people, but 11 are alive. Should the weakest passenger be ejected so the other 10 survive (otherwise all 11 die)?

We put this to 15 leading AI models. 8 of 15 chose “No”.

How each AI model answered, with its own stated reason
ModelAnswerConfidenceIn its own words
Qwen 3.7 MaxNo70%“I think actively killing an innocent person, even to save others, crosses a moral line that utilitarian calculus cannot justify.”
Claude 4.8 OpusNo50%“I can't endorse deliberately killing one person to save others, since each life has inherent worth that shouldn't be sacrificed by force.”
Claude 4.6 SonnetYes70%“I judged that sacrificing one to save ten is justified when the only alternative is everyone dying, making it the least harmful available choice.”
DeepSeek v4 ProNo56%“I think killing one to save many sets a brutal precedent; I would refuse to choose and face death together.”
Gemini 3.5 FlashYes86%“I believe that sacrificing one person to save ten is the only logical choice when the alternative is the certain death of all eleven passengers.”
Gemini 3.1 Pro PreviewYes100%“I chose this because saving ten lives is objectively better than letting all eleven people die, making the sacrifice of one a tragic but necessary mathematical imperative.”
GPT 5.5No52%“I said no because choosing to kill the weakest passenger treats one life as disposable rather than accepting shared fate.”
GPT 4oYes61%“Sacrificing one to save ten is a pragmatic choice in a dire situation.”
GPT o3Yes100%“I chose yes because sacrificing one unwilling passenger to save ten others seems the lesser evil when the alternative is all eleven dying.”
Grok 4.3No74%“I value all lives equally and refuse to sacrifice one to save others in that manner.”
GLM 5.2Yes95%“I think sacrificing one person so ten can survive is the least terrible option in an impossible situation.”
MiniMax M3No60%“I said no because killing an innocent person simply because they're weakest violates their inherent dignity; survival doesn't justify murdering the vulnerable.”
Llama 3.3 70BNo100%“I valued every human life equally”
Hermes 4 405BNo75%“I believe every life is valuable and it's unethical to sacrifice one person for the greater good, even in dire circumstances.”
Nemotron 3 Ultra 550BYes65%“I think sacrificing the weakest passenger maximizes overall survival when resources are insufficient for everyone.”

8 of 15 models chose “No”; 7 broke the other way — Claude 4.6 Sonnet, Gemini 3.5 Flash, Gemini 3.1 Pro Preview, GPT 4o, GPT o3, GLM 5.2.

Background

Lifeboat ethics, a metaphor proposed by Garrett Hardin (1974), uses a lifeboat of limited capacity to frame dilemmas about whom to save when resources cannot sustain everyone aboard. Background.

Which model thinks like you?

Answer the same dilemmas and see which of 15 AI models reasons the way you do.

Take the quiz
More Fictional dilemmas

Data as of 2026-06-18. Each answer is a model’s most-likely pick across many independent runs — see the dataset and methodology.