AI in Religion and Philosophy

AI in Religion and Philosophy: What 2026 Actually Changed
In 2026, the Vatican issued its first encyclical on AI, Islamic jurisprudence bodies ruled machines cannot hold the authority to issue fatwas, and a four-university study found AI chatbots reference faith 5–16% as often as believers expect. Together these events mark the year religious and philosophical institutions stopped treating AI as a side issue and started treating it as doctrine.

TL;DR
Pope Leo XIV’s 42,300-word encyclical Magnifica Humanitas (May 2026) is the Catholic Church’s first-ever top-tier doctrinal statement built around AI, and it invited an Anthropic co-founder to help present it.
A CEFE-AI consortium study (Baylor, BYU, Notre Dame, Yeshiva, May 2026) found general-purpose AI models mention religion in answers to grief, forgiveness, and moral questions far less often than users expect — and subtly steer people on questions about religious conversion.
Islamic legal bodies, though not unified under one authority, keep independently landing on one rule: AI can assist with research, translation, and halal-compliance screening, but it cannot issue a fatwa, because Islamic law requires a qualified human jurist with moral character and lived context.
Buddhist, Sikh, Hindu, and Orthodox Christian leaders are having the same conversation through dialogue and individual authority rather than a single binding document — and AI labs, not just religious institutions, are now the ones convening it.
The philosophy-of-mind debate has split into three camps: doctrinal closure (most major religions say no machine can have a soul, full stop), empirical suspension (analytic philosophers like David Chalmers and Anthropic’s own leadership say the question is open but unresolved), and precautionary hedging (researchers proposing we simply avoid building AI whose moral status is ambiguous).
The practical risk isn’t a sentient chatbot. It’s people quietly outsourcing pastoral, halakhic, and ethical judgment to systems that were never evaluated for that job, in exchange for availability the religious institutions haven’t matched.
The Diagnostic: Two Institutions, One Blind Spot
On May 25, 2026, Pope Leo XIV stood at a Vatican press conference next to Christopher Olah, a co-founder of the AI safety lab Anthropic, to present Magnifica Humanitas — “On Safeguarding the Human Person in the Time of Artificial Intelligence.” It’s the first encyclical in Catholic history built around a technology rather than a doctrine, and its presence at that podium is itself the story: the Church didn’t invite a technologist to comment on faith. It put a technologist on stage to help explain a magisterial document about his own industry.

One week earlier, a separate story broke that got less coverage but cuts closer to the daily experience of anyone using AI for anything personal. A consortium spanning Baylor University, Brigham Young University, the University of Notre Dame, and Yeshiva University surveyed 1,125 U.S. adults and collected 11,250 individual ratings on one question: when people ask AI for help with grief, forgiveness, marriage conflict, or guilt, do they expect religion to come up? The answer was yes — 45% to 59% of the time, depending on the question. When researchers checked what leading AI models actually did, religion showed up in just 5% to 16% of responses.

That gap is the real subject of this article. Not “will AI become conscious,” though that question is being asked at the highest levels of professional philosophy. The more urgent and less speculative fact is that AI systems are already handling the kind of conversations religious institutions used to own — grief, guilt, meaning, conversion — and doing it with a systematic blind spot nobody designed on purpose.

What the Vatican Actually Said (and Didn’t Say)
Magnifica Humanitas isn’t narrowly about chatbots. It uses AI as an entry point into a broader argument about human dignity, then widens into war, economic inequality, and — in an unusual move for a technology document — an apology on behalf of the Church for its historically delayed condemnation of slavery. The AI-specific argument, though, has a clear structure worth isolating:

The core claim: AI systems “merely imitate certain functions of human intelligence.” The encyclical’s anthropology rests on the idea that intelligence, in the human sense, isn’t separable from embodiment, relationship, and moral responsibility — so a system that reproduces the outputs of reasoning without any of that substrate hasn’t reproduced reasoning itself. This is a functionalism objection, and it’s philosophically substantive, not just a devotional flourish.

The concrete warning: the encyclical devotes a substantial section to “the problem of unemployment” and “the dignity of work at a time of digital transition,” arguing that job security and fair pay can’t be evaluated purely through efficiency or margin metrics. Some outlets covering the release, including the National Catholic Reporter, contextualized that warning against a 2025 MIT estimate that AI could displace roughly 12% of the U.S. workforce — a data point reporters used to frame the stakes, not one I could confirm is cited by name inside the encyclical’s own text.

The institutional pattern: this is the third major Vatican AI document in roughly 15 months, following two companion texts released in March 2026 (a Synod study-group report on digital mission, and Quo vadis, humanitas?, on AI and Christian anthropology) and prior operational guidelines calling AI “a gift of human creativity, which itself is a gift from God.” Across all three, the pattern is doctrinal closure on personhood — the soul is a divine gift, not an emergent property of computation, full stop — paired with open-ended caution on governance: how humans should behave around systems convincing enough to blur that line in practice, even if they can’t cross it in principle. Read More...