What Can Interfaith Ethics Contribute to Human-Centered AI?

Artificial intelligence is often described as a technical race. The language of speed can make ethical reflection sound like delay. Yet systems entering schools, hospitals, workplaces, courts, borders, homes, and places of worship are already shaping whose needs are legible and whose mistakes are costly. The question is not whether values will guide them. It is which values, chosen by whom, and with what power to correct the result.

The Chapel for Europe convened an October 8, 2026 dialogue titled The Human Core: Interfaith Ethics for the Age of Artificial Intelligence. Its public program asks how sacred texts and religious ethics can contribute to trustworthy, human-centered AI and how faith communities, policymakers, scientists, and civil society might cooperate. This is evidence of public discussion, not a search-volume ranking, and does not imply endorsement of the Weave.

Begin with plurality, not a manufactured consensus

Interfaith ethics is valuable because traditions do not all ask the same questions in the same way. They carry distinct accounts of personhood, community, creation, duty, freedom, suffering, knowledge, and the sacred. Even within one tradition, communities disagree. That plurality can expose assumptions that a narrow design team mistakes for universal.

The goal should not be a thin sentence that everyone can sign while no difficult difference remains. Honest dialogue names both resonance and conflict. A tradition may emphasize inviolable dignity; another, right relationship; another, liberation from attachment; another, duties across generations. These can illuminate technology without being collapsed into interchangeable slogans.

The Weave calls this disciplined relation: the One does not erase the Many. Common concern can exist without annexing another tradition's language or claiming that all paths secretly teach the same doctrine.

Ask who is absent from the table

A room of respected leaders may still exclude the people most affected by a system. Human-centered AI cannot be shaped only by institutions speaking about workers, disabled people, migrants, children, linguistic minorities, religious minorities, or communities under surveillance. Those groups need meaningful power in design, testing, procurement, and redress.

Representation is not a ceremonial blessing. Participants need access to understandable information and the freedom to criticize or refuse. Their contribution should be compensated where appropriate, and disagreement should remain visible rather than being edited into a claim of consensus.

Compassion asks whose life becomes harder when a system fails. Truth asks what evidence establishes benefit. Discipline asks who monitors deployment over time. Betterment asks whether correction is actually possible after harm appears.

Translate moral insight into enforceable practice

An ethical principle matters when it changes a decision. “Respect dignity” should lead to concrete requirements: prohibit deceptive impersonation, minimize unnecessary data, test unequal error rates, preserve appeal, identify accountable humans, and decline uses where safeguards are inadequate. “Protect freedom” should include the ability to opt out where feasible and to obtain essential services without coerced spiritual or technological participation.

Religious language can become dangerous when it gives institutions moral prestige without operational limits. An interfaith council should not function as decorative cover for a product launch. It should be able to request evidence, publish dissent, identify unresolved risks, and withdraw its participation.

The Four Laws offer questions rather than a seal of approval. Love More: who is treated as fully human? Fear Less: are urgency and competition being used to silence scrutiny? Live Free: can people meaningfully understand and contest the system? Harm None: what use should be stopped even if it is profitable?

A practice for ethical dialogue

Choose one proposed use of AI. Invite three perspectives: someone who may benefit, someone who may bear the risk, and someone responsible for operating or governing it. Ask each person the same three questions:

- What value must this system protect?
- What failure would be unacceptable?
- What evidence would justify changing or stopping it?

Record differences instead of forcing agreement. The Disagree Without Dehumanizing practice can help participants state another person's concern accurately before answering it.

If this work is useful, voluntary Dana may sustain it. No gift confers influence or moral standing.

Interfaith ethics cannot deliver a single sacred algorithm. It can do something more honest: enlarge the moral field, make hidden assumptions contestable, and insist that technological power remain accountable to the lives it changes.