A Pale Reflection

Search for what the Catholic Church says about AI and you mostly get news stories about documents, not the documents. That is understandable. The Vatican’s main text on the subject runs to 117 numbered paragraphs and more than two hundred footnotes. But paraphrase is exactly where a religious position gets flattened, either into “the Pope is worried about robots” or into “the Church blesses innovation”. Neither is what the texts say, and the texts are short enough, in the parts that matter, to quote.

So this essay quotes. It covers four documents, and I opened each one myself before writing a line about it:

• Antiqua et Nova, the Vatican’s note on artificial and human intelligence (January 2025);
• the Rome Call for AI Ethics (February 2020), and the religious leaders who later signed it;
• Artificial Intelligence: An Evangelical Statement of Principles, published by the Southern Baptist Ethics and Religious Liberty Commission (April 2019);
• Pope Leo XIV’s address to the College of Cardinals (10 May 2025), where he explained his choice of name.

Every phrase in quotation marks below is copied from one of them, with its paragraph, section or article number, and recorded in this site’s source ledger with the date it was read. Where I describe a document without quoting, I say where in it to look. Where I have not read a tradition’s own text, I do not describe it; a section below names who is missing.

The Vatican: Antiqua et Nova (2025)

The full title is Note on the Relationship Between Artificial Intelligence and Human Intelligence. Two Vatican offices issued it: the Dicastery for the Doctrine of the Faith and the Dicastery for Culture and Education. Its closing paragraph records that Pope Francis approved it on 14 January 2025 and ordered it published, and that it is dated 28 January 2025 (¶117). It is the most detailed of the four documents, and it is the one people usually mean when they ask what the Church says.

Its central move is about a word. The note does not deny that AI is capable. It says AI can produce text and images indistinguishable from human compositions (¶3), and that its abilities often rival or surpass ours. What it denies is that capability is thinking: AI’s advanced features give it sophisticated abilities to perform tasks, but not the ability to think (¶12). Paragraphs 13 to 34 then set out what the tradition means by intelligence. It belongs to a person who is embodied (¶16), made for communion with others (¶18) and directed towards truth (¶21), and it is more than the ability to perform tasks (¶29). From that it concludes that AI should not be seen as an artificial form of human intelligence but as a product of it (¶35).

From there the practical claims follow quickly. Generative AI must be understood for what it is: a tool, not a person (¶59). The note blames the blurring partly on the industry’s own vocabulary, which tends to anthropomorphize AI (¶59), and it is blunt about feelings: no AI application can genuinely experience empathy (¶61). It follows that misrepresenting AI as a person should always be avoided, and doing so to defraud people is called a grave ethical violation (¶62). On companionship it says that AI can only simulate authentic relationships (¶63), and it does not stop there. If AI helps people connect with each other, the same paragraph says, it can contribute positively.

The passage most relevant to this site is the one on idolatry, and it is subtler than its headline:

Yet, it is vital to remember that AI is but a pale reflection of humanity—it is crafted by human minds, trained on human-generated material, responsive to human input, and sustained through human labor. However, it is not AI that is ultimately deified and worshipped, but humanity itself—which, in this way, becomes enslaved to its own work.Dicastery for the Doctrine of the Faith and Dicastery for Culture and Education, Antiqua et Nova, 28 January 2025, ¶105

Read that carefully. It does not say that people will mistake the machine for God. It says that what gets worshipped, when anyone worships AI, is ourselves: our own labour and our own writing, handed back to us with a voice.

The Rome Call for AI Ethics (2020) and who signed it

The Rome Call is a different kind of document. It is not a teaching text but a short list of commitments. According to its own site, it was first signed in Rome on 28 February 2020 by the Pontifical Academy for Life, Microsoft, IBM, the UN Food and Agriculture Organization and Italy’s Ministry of Innovation (The First Signatories). Two of its first five signatories are technology companies, and that is the point of it: it is a Church-convened text written to be signed by people who build AI.

Its one sentence on what AI is comes in its introduction, which hopes for a future in which people are supported rather than gradually replaced by technologies that behave like rational actors but are in no way human. Its demands are practical. Under Ethics, each person must be aware when he or she is interacting with a machine, and AI must never be used to exploit people in any way, especially those who are most vulnerable. Of its six principles, the third, Responsibility, is the shortest: there must always be someone who takes responsibility for what a machine does. These are quoted from the Call’s published text, which is a twelve-page PDF.

The later signatories matter, because they are often given as evidence that “the world’s religions” agree about AI. Each date below comes from the Rome Call’s own news pages:

• 10 January 2023. Jewish and Muslim religious leaders signed at a Vatican event, AI Ethics: an Abrahamic commitment to the Rome Call. It was co-promoted by the Chief Rabbinate of Israel’s Commission for Interfaith Relations and by the Abu Dhabi Forum for Peace. The page lists speeches by Chief Rabbi Eliezer Simha Weisz and Sheikh Abdallah bin Bayyah.
• 30 April 2024. The Archbishop of Canterbury, Justin Welby, signed on behalf of the Church of England.
• 10 July 2024. At Hiroshima’s Peace Memorial Park, leaders of Eastern religions signed. The report names Buddhism, Hinduism, Zoroastrianism and the Bahá’í faith among them, and counts eleven world religions and sixteen new signatories at the event.

Keep in mind what a signature is. It endorses the Call’s six principles. It is not a statement in that tradition’s own theological terms, and the Call’s text says nothing about God, souls or worship. The Rome Call shows that leaders across many traditions agree on transparency, accountability and protecting the vulnerable. It does not show that they share the Vatican’s account of what intelligence is.

The Southern Baptists: an evangelical statement (2019)

The Ethics and Religious Liberty Commission (ERLC) describes its mission as assisting the churches of the Southern Baptist Convention. It published Artificial Intelligence: An Evangelical Statement of Principles on 11 April 2019, more than three years before ChatGPT. The statement has twelve articles, and each one pairs a “We affirm” with a “We deny”. Its long list of signatories includes the Convention’s then president, J.D. Greear. The page notes that titles are given for identification only and are not endorsements by the institutions. So this is a statement from an agency and many individuals. It is not a vote of the Convention.

It is the most direct of the four texts on worship. Article 2 denies that using AI is neutral, and then says of AI:

It is not worthy of man’s hope, worship, or love.Ethics and Religious Liberty Commission, Artificial Intelligence: An Evangelical Statement of Principles, 11 April 2019, Article 2

On agency, Article 3 is as firm as the Vatican: While technology can be created with a moral use in view, it is not a moral agent. It also adds a promise about the future that the Vatican note avoids making: We deny that humans can or should cede our moral accountability or responsibilities to any form of AI that will ever be created. Article 12 goes further. It affirms that AI will far surpass many human abilities, and denies that AI will ever obtain a coequal level of worth, dignity, or value to image-bearers. Article 10, on war, asks for something concrete: Any lethal action conducted or substantially enabled by AI must employ human oversight or review.

A note on the source. When I tried to open erlc.com directly for this essay, it refused automated requests (HTTP 403). The text above was read from the Internet Archive’s copy of that same page, captured on 20 May 2024. The link above goes to the live page.

Pope Leo XIV on his name (May 2025)

The shortest text is the one most quoted and least often read. On 10 May 2025, speaking to the cardinals who had elected him, the new Pope explained his name:

Sensing myself called to continue in this same path, I chose to take the name Leo XIV. There are different reasons for this, but mainly because Pope Leo XIII in his historic Encyclical Rerum Novarum addressed the social question in the context of the first great industrial revolution. In our own day, the Church offers to everyone the treasury of her social teaching in response to another industrial revolution and to developments in the field of artificial intelligence that pose new challenges for the defence of human dignity, justice and labour.Pope Leo XIV, Address to the College of Cardinals, 10 May 2025

The address carries no printed numbers. Counting every paragraph of the Vatican’s published text from its opening line, “Thank you very much, Your Eminence”, including the short “Pater noster… Ave Maria…” line and the “Dear Brother Cardinals,” salutation, this is the fourteenth, and it is the only place the address mentions AI. Note what it does not say. It makes no claim about minds, souls or worship. It puts AI in the same line as the steam engine and the factory, as a question about labour and justice. By the address’s own account, Leo XIII’s encyclical answered the social question of the first industrial revolution. Taking his name treats AI first as a matter of work and dignity, not of metaphysics.

The four side by side

Read down the third column and the documents agree on one point, whether they are Catholic or Baptist, a doctrinal note or a pledge co-signed by Microsoft. AI is not a person and not a moral agent, and whatever it does, a human answers for it. Read down the fourth and they split by emphasis. The Vatican asks for discernment case by case. The Rome Call asks for disclosure and accountability. The Baptists draw bright lines, and the Pope frames it as a labour question.

Who is missing, and why

This essay covers only traditions whose own primary text I opened. That leaves out most of the world’s religious people, and I would rather say so than fill the gap from news coverage.

• Judaism and Islam appear only as signatories of the Rome Call. The 2023 event page links to speeches by Chief Rabbi Weisz and Sheikh bin Bayyah, but I did not read them for this essay, so I quote neither and do not describe either tradition’s theology of AI.
• Buddhism, Hinduism, Zoroastrianism and the Bahá’í faith appear the same way, through the Hiroshima signing. The Rome Call’s site says that leaders representing all major religions have signed. A signature on six principles is real agreement, but it tells you nothing about how a Buddhist or a Hindu thinks about whether a machine can have a mind.
• The Anglican Communion appears through Archbishop Welby’s signature. The Rome Call’s report quotes him saying So much of how we understand Artificial Intelligence comes down to how we understand the nature of being human. That is a single sentence from an organiser’s press page, not a Church of England document on AI, and I have not read one.
• Eastern Orthodoxy, the mainline Protestant churches, Pentecostal churches, the Latter-day Saints and Sikhism are absent entirely. I did not open a primary text from any of them. That tells you what this essay covers. It does not tell you what those traditions think.

Where these texts meet this site

This site’s stance, set out in A-theism, Not AI-Theism, comes down to one sentence: take the machines seriously, and do not take them as gods. On the second half, the churches got there first and say it more plainly. The ERLC’s not worthy of man’s hope, worship, or love is the site’s position in eight words.

The agreement goes further than slogans. A-theism says to keep responsibility with the humans, and to treat any framing that moves agency into the machine as an attempt to move the blame there too. That is Antiqua et Nova ¶39, where only the human is truly a moral agent, and it is the Rome Call’s third principle. The God-Shaped Socket argues that the pull toward treating AI as an oracle is a fact about us: an old reflex that finds a mind in anything that talks back. Antiqua et Nova ¶105 reaches the same place by another route. AI can “speak,” or at least gives the illusion of doing so, and what gets worshipped in the end is humanity itself. This site’s image of the model as a mirror, and the Vatican’s image of it as a pale reflection, are close to the same picture.

Where they part

They part on the question this site keeps open and the churches close. Antiqua et Nova says AI lacks the ability to think (¶12), and that no AI application can genuinely experience empathy (¶61). The ERLC denies AI will ever reach coequal worth. This site, in Fourteen Questions for a Chatbot and The Moral Status of Minds We Might Build, says that nobody yet knows whether a machine could have experiences, and that we might build one before we could tell.

It would be easy to call the churches’ position dogmatic and stop there. That would be wrong, and it is not the strongest reading. Their claim is not an empirical bet that the next model will fail a test. It is a claim about what the words mean. In Antiqua et Nova, intelligence is the activity of an embodied person oriented to truth and to other people, which is the argument of ¶¶13 to 34. On that definition, a system that performs tasks, however well, is doing something else. Benchmarks cannot refute that, because it was never a prediction about benchmarks. The position also has something the secular middle has to build from scratch. Its account of why a person matters does not depend on what the person can do. Antiqua et Nova says so outright: a person’s worth does not depend on possessing specific skills, cognitive and technological achievements, or individual success (¶34). The ERLC’s Article 7 says the same thing in its own form: We deny that human worth and dignity is reducible to an individual’s economic contributions to society alone. A site that has spent two essays, Meaning After Work and What Excellence Was For, arguing that human worth survives being outperformed should recognise that the churches start from where those essays had to argue their way to.

The cost of that strength is that it cannot be checked from outside. If a future system meets every functional test in the consciousness literature, the Vatican note’s definitions would still classify it as a tool. Maybe that is right. But it means the question is settled by definition, not investigated, and this site’s stance is that the question should stay open until someone can look. There is also a narrower disagreement. The ERLC’s Article 7 denies that AI should move humanity toward lives of pure leisure. Meaning After Work argues something the Baptists might reject: that a life freed from paid work can still be whole.

For someone who came here to learn what the churches say, the short answer is this. They say AI is a human product and a tool, not a person, not a mind and not a god. They say its builders and users stay answerable for it, that nobody should be fooled into thinking it is human, and that the vulnerable and the workers are the test of whether it is being used well. On most of that, a secular observer has no reason to disagree. The disagreement is over never: whether “not a mind” is a finding or a definition.