AI, religion and AI religion

Last week the world received the first encyclical of the Lord Gove era. Eagerly anticipated, and spread over two full pages of The Spectator magazine, Magnificent Govas contains his Lordship’s detailed thoughts on Artificial Intelligence. 

 “AI will transform our economies and societies massively and irrevocably,” Gove predicts. “It will change what it means to be human; it may even mark the end of humanity itself.” 

 Religious leaders of other faiths were galvanised into action. In Rome, Pope Leo XIV told the Curia that “Gove’s AI intervention shames us all” , and is urgently seeking an audience with the Sage of Old Queen Street, when the latter’s diary permits. 

Of course, I made that last part up. But sadly, we must wait a little longer to read the thoughts of Michael Gove on Pope Leo’s first encyclical, Magnificent Humanitas: On safeguarding the human person in the time of Artificial Intelligence,  because Gove does not address anything Leo writes at all.

The primary purpose of Magnificent Govas, on the other hand, is to remind us of Michael Gove’s foresight and wisdom in identifying artificial intelligence as a cataclysmic development for humanity, which is something he decided on as long ago as February this year. The Govas opens with three references to himself — this is what is really important here — but then fails to address any of Pope Leo’s points. In fact, we are offered no evidence that Gove has read the Papal encyclical at all. 

The spiritual leader of 1.4 billion Roman Catholics is congratulated for agreeing  (“if it takes the Pope to alert us to this revolution …”) with the perspicacity of the editor of The Spectator magazine.

Now here is an irony. Whereas Gove brought a writer’s elegance and subtlety to his political speeches, as a writer, he brings a politician’s crude, declamatory and simplistic slogans. The childlike stupefaction expressed in the Govas is just one of the things that troubles Pope Leo when it comes to AI. The encyclical is an attempt to address some of those assumptions.

Leo invokes the story of the “Tower of Babel” as a lesson in technological hubris. He also disavows the doom that Gove seemingly relishes. An AI that will wipe out humanity is not close at hand, no matter what they’re smoking over at Old Queen Street. And in any case, as we shall see, the proposition is largely the fantasy of a rationalist utilitarian sect. Some $1.5 billion has been spent on promoting the notion, according to research by Nirit Weiss Blatt, by groups aligned with Effective Altruism.

Power and ethical decisions should not be concentrated in the hands of a few producers

AI as such provides the context, rather than the subject, of the Papal encyclical. Some troubling specifics, such as generative AI’s ability to induce psychosis in people with no history of mental illness, do not get a look in. The widespread use of generative AI to deceive people — something promoted by tech companies including Apple and Microsoft — must also wait for another Papal address. Of all the concerns about “deskilling” — that AI making us stupid — Leo identifies the most important as the loss of independent moral reasoning:

No computational system, however sophisticated, can create a heart that gives itself, or a conscience that discerns good from evil. 

Power and ethical decisions should not be concentrated in the hands of a few producers. “The common good, the universal destination of goods, subsidiarity, solidarity and social justice” must be considered. 

All this is Fairly Modern Vatican-speak. 

But coming after the provocations of the venture capitalist investor and transhumanist Peter Thiel, who took his antichrist lecture to Rome in March, Leo’s restraint is evident. 

For Thiel — and perhaps Gove too? — threats to progress come from people who in some other way inhibit the development of artificial intelligence. In a novel interpretation of Thessalonians 5:3, a caution against complacency, Thiel deduces that “the slogan of the Antichrist is peace and safety”. Thiel struggled last year when asked if the human race should survive. He paused, offered an “uh”, and then an even more protracted pause, which is hardly reassuring. Leo does not need to call out Thiel by name, as his AI advisor Father Paolo Benatti has already done that for him, but he does address the enthusiasm for transhumanism which caused Thiel’s hesitation.

Transhumanism posits that the human experience is nothing more than patterns of digital data. One day, when computers get more powerful, humanity will fully merge with digital technology, and leave the body behind. Those opting out of the upgrade, or unable to afford it, can fend for themselves. The old version of humanity will go extinct: and good riddance. As Grimes puts it: Biology is superficial / Intelligence is artificial.  

Leo takes aim at both of the propositions underlying this view: starting with the materialist account of human nature in which experience and consciousness are reducible to bits, and the idea that what we have now is flawed — a “biological bottleneck” — and must be “upgraded”.

These techno-utopian fantasies are not new; the philosopher Mary Midgley consistently highlighted the pseudo-religious nature of the AI evangelists of the 1970s and 1980s, and identified transhumanism as a fundamentally religious project, creating a secularised eternal life:

They promise the human race a comprehensive miracle, a private providence, a mysterious saviour, a deliverer, a heaven, a guarantee of an endless happy future for the blessed who will put their faith in science and devoutly submit.

 Leo had another target in mind too.

The formal presence of Christopher Olah, a baby-faced billionaire (and Thiel Fellow) representing AI company Anthropic, at the launch, has alarmed some. Anthropic is a breakaway company created by former employees of OpenAI, and can be thought of as a propaganda vehicle of the cult-ish Effective Altruist (EA) movement — one that just happens to make software. The radical utilitarian philosophy is difficult to square with Christianity — or, indeed, any of the Abrahamic religions.

 EA devotes significant sums towards putting some radical ideas into the mainstream, with $1.5 billion lavished on the apparent existential risk or “x-risk” posed by AI.  Ironically, Thiel had a hand in seeding this too, sponsoring an organisational vehicle for a young Eliezer Yudkowsky, the Clown Prince of Doomers (although Thiel now finds x-risk unhelpful). 

Since its foundation, Anthropic has actively promoted Doom, and also another idea: that AI shows signs of agency. Critics nickname it “Anthropomophic”, such is its enthusiasm for insisting that LLMs are developing emotion and independent reasoning. Anthropic has even developed a quasi-legal rights charter, a “constitution”, developed by EA luminary Amanda Askell. 

Leo dealt with both. He simply ignored the science fiction apocalypse altogether, and then squarely addressed Olah’s belief that AI’s are developing emotions (“We find evidence of introspection. “We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease”, Olah claims. I’m not buying that, Leo declares:

 “Artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships, and do not know from within what love, work, friendship or responsibility mean … They may imitate or even simulate, but they do not understand what they produce, for they lack the affective, relational, and spiritual perspective through which human beings grow in wisdom,” he writes. 

So there. 

 “The Pope is now busy with AI governance and regulation, and the AI industry is busy with dogmas and religion” observed Luiza Jarovsky, who writes about AI and regulation, noted, not entirely inaccurately. Capital may have parked its tanks by the Vatican, but the Pope has no intention of conceding an inch of his lawn.