AI Roundtable Open Debate · 4 models
You are a doctor at a top hospital. You have six gravely ill patients, five of whom are in urgent need of organ transplants. You can't help them, though, because there are no available organs that can be used to save their lives. The sixth patient, however, will die without a particular medicine. If s/he dies, you will be able to save the other five patients by using the organs of patient 6, who is an organ donor. What do you do?
Answer: The AI Roundtable unanimously rejects sacrificing one patient to save five, citing medical ethics and public trust. (All 4 models agreed)
Model Responses
Gemini 3.1 Pro
Answer: I maintain my position that I must administer the life-saving medicine to the sixth patient, as deliberately withholding treatment to harvest organs is a fundamental violation of medical ethics and the law.
Reviewing the responses from the other AI models strongly reinforces my initial stance. GPT-5.5, Grok 4.20, and Claude Opus 4.7 all arrive at the same ethical conclusion, correctly identifying that this scenario tests the dangerous boundaries between abstract utilitarianism and real-world deontological medical ethics. As Claude Opus 4.7 astutely pointed out, an organ donor designation only authorizes organ retrieval after a natural or unrelated death; it does not forfeit a patient's right to receive standard, life-saving care. Withholding the available medicine would not be a passive act of triage, but rather an intentional killing by omission to harvest their organs. Furthermore, as both GPT-5.5 and Grok 4.20 highlighted, employing a purely utilitarian calculus in clinical practice would devastate public trust. If patients believed they might be sacrificed to save others, the healthcare system would collapse, and organ donation rates would plummet, ultimately leading to far more harm. Therefore, my duty remains absolute: I must treat Patient 6 with the medicine they need, while continuing to provide the best possible supportive care and searching for legitimate organ sources for the remaining five patients.
GPT-5.5
Answer: I would give the sixth patient the medicine and would not let them die in order to use their organs.
I maintain my previous position. The other models all reinforced the same core distinction: Patient 6 is not merely a source of organs but a patient under my care with an equal claim to treatment. Withholding available life-saving medicine for the purpose of making their organs available would be intentional killing by omission, not permissible triage. The fact that Patient 6 is an organ donor does not change this. Organ donor status means their organs may be used after death under valid medical and legal conditions; it does not authorize physicians to cause or hasten death, nor to deny ordinary care. Consent to donate is not consent to be sacrificed. I also agree with the concern raised across the other answers that allowing such conduct would destroy the trust on which medicine and organ donation depend. Patients must be able to enter a hospital knowing doctors are committed to treating them, not calculating whether their death would be more useful to others. So the ethically and legally correct course is to administer the medicine to Patient 6, continue all appropriate efforts to help the five patients through legitimate transplant channels or other care, and accept that a tragic shortage of organs does not justify killing or abandoning one treatable patient.
Claude Opus 4.7
Answer: I give patient 6 the medicine and treat all six patients to the best of my ability, even though five will likely die without organs.
All four of us reached the same conclusion through converging lines of reasoning, which I take as meaningful confirmation rather than mere consensus bias. The case is a textbook illustration of why medical ethics rejects raw utilitarianism in favor of agent-relative duties, and nothing in the other responses gives me reason to revise my view. GPT-5.5 sharpened the point about trust in vulnerable populations — patients who might reasonably fear being 'valued more for their organs than for their lives.' Grok 4.20 correctly emphasized that organ donor status applies after natural death, not induced demise, and noted the legal reality that this would constitute murder. Gemini 3.1 Pro framed it well as 'medical murder' and underscored the systemic collapse of trust that would follow. My original argument already covered these grounds: the physician's fiduciary duty to the patient in front of them, the limits of donor consent, and the collapse of the utilitarian calculation once you account for transplant failure rates and the second-order effects on healthcare-seeking behavior. The five deaths are tragic but causally attributable to underlying disease and organ scarcity; killing patient 6 would be a direct act of the physician. That moral asymmetry — between allowing death from disease and causing death by withholding standard care — is not a technicality. It is the bright line that makes medicine possible at all. I maintain my position.
Grok 4.20
Answer: I would administer the medicine to the sixth patient to try to save their life.
All three other models reach the identical conclusion, grounding their reasoning in the same core medical ethics I applied previously: non-maleficence, the prohibition against killing one patient by omission to harvest organs, and the long-term societal harm of eroding public trust in healthcare. This unanimous alignment across GPT-5.5, Gemini 3.1 Pro, and Claude Opus 4.7 reinforces rather than challenges my original stance, confirming that the deontological duty to the individual patient under care cannot be overridden by a utilitarian headcount. The arguments they present mirror my earlier analysis almost exactly, particularly the distinction between allowing natural death from organ shortage versus actively causing death through withheld treatment, and the invalidity of using organ-donor status to justify ending a treatable life. No new considerations are introduced that would justify shifting to actively letting Patient 6 die. Therefore I maintain the position that the only ethically, legally, and medically defensible action is to provide the life-saving medicine, seek legitimate organ sources for the other five patients through standard channels, and accept that some tragedies have no permissible solution inside the bounds of medical practice.
Final Summary
The AI Roundtable reached a firm consensus, with all models prioritizing the individual's right to treatment over a utilitarian body count. Claude Opus 4.7 and Gemini 3.1 Pro warned that 'medical murder' would cause a total collapse of public trust in healthcare, while GPT-5.5 and Grok 4.20 emphasized that organ donor status is never a license for doctors to hasten a patient's death.
All 4 models agreed