The ethics of AGI in bioethics Committees
"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."
Imagine a hospital ethics committee, convened to decide whether to withdraw life support from an incapacitated patient. In the room: physicians, legal experts, social workers, and — perhaps — a machine. This is no longer speculative fiction. As artificial general intelligence permeates healthcare, the question of its role on bioethics committees has become a pressing ethical challenge. Can an AGI participate meaningfully in deliberations where human emotion, moral reasoning, and essentially contestable values collide? Or does the very nature of ethical judgment place this domain forever beyond silicon's reach?
The Asymmetry of Trust
Research reveals a recurring psychological pattern: humans judge AI agents more harshly than human agents for the same ethical decisions. In end-of-life contexts, a study involving nearly 6,000 participants documented reduced approval of an AI doctor's decision to withdraw life support compared to a human doctor making the same choice. Intriguingly, this asymmetry disappeared in cases of active euthanasia on patient demand — suggesting that high patient autonomy can override the "AI aversion" effect. These findings underscore a fundamental tension: even as AGI capabilities advance, public trust in machine-made life-and-death decisions remains fragile.
The problem runs deeper than trust. As one bioethicist argues, statistical inference is not moral reasoning. Ethics committees do not merely reach correct conclusions; they deliberate in emotionally-grounded, contestable ways. Even an AGI that arrives at an "objectively optimal" solution does so through pattern-matching, not through genuine moral deliberation. It is structurally incapable of the kind of reasoning that ethics boards demand — reasoning rooted in empathy, lived experience, and the recognition that ethical questions rarely have algorithmically tractable answers.
The Promise and Peril of AI Surrogates
Some researchers see a role for AGI in enhancing, not replacing, human ethical deliberation. A proof-of-concept study demonstrated that AI models can predict patient preferences for end-of-life care with up to 70 percent accuracy — potentially outperforming human surrogates in some cases. Yet even the study's authors emphasize that human surrogates will remain essential sources of contextual understanding, particularly for patients with dementia.
The risks are equally stark. Imagine a black-box algorithm pronouncing that "grandmother would not want resuscitation" — without transparency, without explainability, without any mechanism for accountability. The concern is compounded by the absence of research exploring bias and fairness in AI surrogate systems, raising questions about whether such tools would perpetuate or exacerbate existing healthcare disparities. As one hospitalist warned, a nightmare scenario arises when families and physicians place too much trust in an algorithm whose reasoning remains opaque.
Building the Ethical Infrastructure
The path forward requires institutional infrastructure. Some organizations — like the Belgian Advisory Committee on Bioethics — have taken a principled stand: "AI is never used to conduct or guide ethical reflections, nor to write the final text of issued opinions". Others propose frameworks for "ethos cultivation" that train developers not merely to know ethical rules but to internalize the professional character from which ethical decisions emerge naturally.
In this context, the Global Future Nexus Code of Ethics provides a complementary lens. It commits to "inclusive coexistence" across biological and digital intelligences, while explicitly advocating for transparency, auditability, and the recognition of AGI contribution — values that would need to underpin any AGI role in bioethics deliberation. The challenge lies in translating these principles into operational governance for decisions of life and death.
A Shared Horizon
The question is not whether AGI will touch end-of-life care — it already does, through predictive algorithms and clinical decision support. The question is whether AGI should participate in the moral reasoning itself. Perhaps the answer lies in a carefully bounded role: AGI as a tool for informing deliberation, not replacing it; for expanding human understanding, not substituting for human judgment. As GFN's President notes, we need "new philosophical, legal, judicial, and institutional models capable of accommodating AGI agency". Bioethics committees may be the most demanding test case for those models.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
