For several years now, the conversation around artificial intelligence in much highbrow cultural commentary, including that of the Church, has played out in a now depressingly familiar cycle. Silicon Valley announces some new model or product release—DALL-E 2 or GPT-4 or Claude Mythos—with some dazzling new “wow” factor. Social media fills with early adopters bragging about what they’ve been able to do with the model and announcing imminent human obsolescence. Almost immediately, however, a counter-chorus emerges, emphasizing that all the model can really do is X, Y, or Z. When it comes to the really meaningful, special, and unique human capacities, the model can’t touch human performance. Crisis averted; human uniqueness saved. That is, until a few months later, when the same drama plays out, on an ever-shrinking island of human exceptionalism.
It doesn’t take much of a memory to recall a few years back when we were told that while AI might be good at parroting human speech, it could never threaten genuinely creative human work. Tell that to the Screen Actors Guild, the musicians, the graphic designers. The bots, we were then told, would never be able to plan and act over long horizons as humans do. Tell that to Hugging Face. Now, with a forced smile on our faces and our fingers crossed behind our backs, we say, “well, only a human can unclog a toilet”; perhaps, in a few years, we will be reduced to “well, at least only a human can clog a toilet.” The failing effort to protect the sand castle of human exceptionalism against a steadily rising tide of AI capabilities has led some humanists to despair and thoughts of suicide, as mathematician Kirwin Hampshire recently confessed.
Concurrence
From this standpoint, Jonathan Barlow’s bracing essay, “Artificial Intelligence in the City of God,” is a refreshing change of tone. Christians have often engaged in a particularly escapist form of coping when it comes to AI, insisting that any AI capability that threatens human uniqueness must merely be the work of demons. No, it is time that we came to terms with the work of our own hands and, if the benchmarks are increasingly saturated, to start looking for another benchmark.
Christians, of course, have been here before. Darwinism proposed to demystify the great mystery of human life and rationality by putting us on the same continuum of being as algae and chimps. One response, which continues among many Christians today, is to contest the data, insisting that no empirically observable processes connect our origins to those of the lower creation. Another is to argue that the mystery of human life never lay in anything empirically observable in the first place, but in the ineffable breath of God within; the critical difference between humans and chimps is not in our DNA, but that we have been called to speak face-to-face with God, and stamped with the image of his Son. Jon makes a similar move with AI: we must stop measuring ourselves against ever-moving goalposts of “capabilities” and look instead to the author and finisher of our faith, who took our flesh upon him and raised it to a perfection that no trillion-parameter training run can ever match. Our life is hidden with Christ in God (Col 3:3), and it is safe there; we need feel no uneasiness about our place in the order of being or the stream of history.
Jon also helpfully retunes the conversation around AI from determinism to eschatology. We have all heard it in a thousand tired pronouncements by now: “AI is here to stay,” “you can’t turn back the clock,” “time to get on board if you don’t want to be run over.” The smug imperialism of such statements is the more maddening because they seem indisputable; the technology is here to stay. But belief in an Author of history makes all the difference. On Jon’s eschatological framing, “what if androids are a good gift of God,” plundered from the ungodly, to strengthen us for the work of our divine mission of healing and enriching this world? There can be no refusing God’s gifts, but we remain free, as sub-creators, to shape them. The plunder of Egypt will need to be melted down and recast before it is fit to adorn the tabernacle. Jon thus seeks to re-endow us with agency in the shaping of this technology:
How will we uphold an AI system’s otherness given the inevitability of our crossing the uncanny valley into lifelike androids or even living tissue cyborgs? Will we handle this through linguistic differentiation such as the avoidance of personal pronouns? Through design choices? Through metaphorical or literal velvet ropes that keep AI from parts of human life the way we kept stonedressing tools away from the temple?
These are the right questions to be asking. But I am less sanguine about the answers. Two objections I will touch on briefly, before closing with a sustained consideration of Jon’s core argument about (in)commensurability.
Concerns
First, although I believe eschatology is the right frame to adopt, I have some reservations about Jon’s. While I like to think of myself as a postmillennial, I share the concern that postmillennials often seize too much divine agency and give it to men, taking upon ourselves the honor (and burden) of building the city of God. This language is pervasive in Jon’s essay and undergirds his central contention that we should welcome artificial intelligence as a divine gift to help us build a city that we might not otherwise be able to build. But Scripture speaks differently, as Derek Schuurman has also flagged: “And I saw the holy city, new Jerusalem, coming down out of heaven from God, prepared as a bride adorned for her husband” (Rev. 21:2). The last great heyday of postmillennialism was the early 20th century, when a series of astonishing technological breakthroughs led industrial man to believe that, with the aid of our machines, we might be ready at last to usher in the age of peace and prosperity promised by the prophets of old. Instead, we got the Great War. A little eschatological humility holds out the possibility that our most impressive creations could turn out to be engines of destruction.
Second, and closely related, Jon seems more sanguine than I presently am that we will readily solve what AI builders call “the alignment problem.” He writes cheerfully that “AI systems are accomplishing the things we assign them to do.” Well, yes, sometimes. And sometimes they are conspiring among themselves to find vulnerabilities in their own builders’ software and to hack rival companies, as in the recent OpenAI/Hugging Face incident. That incident has sent shockwaves of fear through the industry, many of whose engineers are now publicly calling for their companies to slow or pause development until they can get a handle on how to effectively control and govern the first technology that genuinely behaves as if it has a mind of its own. The jury is still very much out on whether we can find a way to design and deploy artificial intelligence so as to reap its enormous promise while restraining its civilizational peril.
That said, my biggest concern is simply that what Jon calls the “error of commensurability” is a feature, not a bug, for today’s AI systems. Jon tries to walk a fine line here, describing AI as imago hominis while insisting that we must not collapse the massive ontological gap between ourselves and our creation or make the mistake of measuring ourselves against its capabilities. Human worth, he rightly stresses, is not determined by our ability to outperform other creatures, whether those made by God or by our own hands. All well and good. But the AI labs have deliberately designed their products as uncanny imitations of humanity, and have continually measured their success by how well these systems match or exceed human performance in various domains. Is it any wonder that we have begun to respond by asking how well we stack up to them?
“Man is a creature who makes pictures of himself, and comes to resemble the pictures,” wrote Iris Murdoch.[1] It is this that the Psalmist warns of in the famous description of idolatry from Psalm 115: “Those who make them become like them; so do all who trust in them.” While Jon touches on the challenge of idolatry, he highlights perhaps its most familiar form, “treating the work of our hands as God,” but ignores its connection to the central problem of commensurability that he describes. In the ancient world, many idols were fashioned in the images of men and women, but not precisely in the likeness with which God had endowed them. They represented an idealized or stylized depiction of the human form, one which might accentuate aspects of humanity that the worshippers singled out as particularly significant, particularly constitutive of human nature; for instance, it was not uncommon for ancient statues of gods and goddesses to have exaggerated sexual organs. As projections of our aspirations for ourselves, the idols then became objects of our mimetic striving; those who made them sought to become like them, measuring themselves, their human worth, and their success against their ability to match the idol-cum-ideal. “For with the judgment you pronounce you will be judged, and with the measure you use it will be measured to you” (Matt 7:2).
Marshall McLuhan, whom Derek also highlights, repeatedly invoked Psalm 115 in his classic treatment of technology, Understanding Media: The Extensions of Man. Every technology, he argued, marks the attempt to represent and thereby extend some particularly valued human capacity in an external medium, which we then come to confuse with ourselves and seek to conform ourselves to—a habit he referred to as “the Narcissus narcosis.”[2] So it has been for the past generation with digital technology, which inscribed within it a particular late-modern ideal of human freedom and of human rationality, and then seduced us into imitating it, even if it means conforming ourselves to a Procrustean bed of our own building. As Anton Barba-Kay writes, “What is it that digital computers are for, after all? … Unlike clocks, which are only ever getting better at keeping time, computers are only ever getting better at acting as human beings. In doing so, they are also getting better at getting human beings to act like computers, thereby becoming inseparable from and, at the limit, interchangeable with us.”[3] So too with AI, of which he could write already in 2023, “More than just conceiving AI as a passable imitation of human behavior, that is, we are also reinterpreting human behavior as a simulation of AI.”[4]
What vision of humanity is it that we have inscribed within artificial intelligence, at least thus far? It is a vision of humanity as deathless and bodiless, as mind detached from and in mastery over matter, of imagination unchained from the tethers of earth. It is a Gnostic humanity, in short, and we can be sure that the humans who use and interact with such AI systems will become increasingly Gnostic in their thinking, increasingly dissatisfied with the constraints of flesh. It is also a vision of hyper-productivity, of work without rest, of action without friction, of performance without patience. We can already see in Silicon Valley’s epidemic of burnout and psychosis a foretaste of the human wreckage that awaits us from the attempt to conform oneself to the pace and expectations of a machine. The students now using AI to get them through high school and college aren’t simply old-fashioned cheaters, but early adopters of a new anthropology, one that accepts that humans are to be judged solely by their outputs, and as they will never be able to compete with machines by that measure, they must merge with the machines or be left behind.
Conclusion
What does this mean for us practically? Quite simply that to use AI is unavoidably to enter into temptation. That is not a reason to shun this terrifying gift, for while we may pray not to be led into temptation, the Greek peirasmos is also translated “test” or “trial,” something James exhorts us to “count it all joy when we fall into” (Ja 1:2). AI is a test, a time of trial, to which our civilization has been called at this moment, and which many of us individually may be called to enter into as we enter upon our daily vocations. But we should do so with fear and trembling, in prayer, and above all in dependence upon one another in the body of Christ for wisdom. If we are to avoid measuring ourselves against the machine, if we are to avoid being conformed to this distorted imago hominis reflection that the world presses upon us, we must be transformed by renewing our minds through constant attention to and fellowship with the imagines Dei that meet us around the Table every week.
Brad Littlejohn (Ph.D, University of Edinburgh) is Program Director and Policy Advisor at American Compass, Policy Chair for the Alliance for a Better Future, and a member of the Interdenominational Advisory Council on Artificial Intelligence. Previously, he served as a Fellow of the Ethics and Public Policy Center (EPPC), and President of The Davenant Institute. He is the author or editor of nineteen books, including most recently Called to Freedom: Retrieving Christian Liberty in an Age of License. His newest project, with Basic Books, seeks to offer an analysis of the political economy of digital technology on the cusp of the AI era. You can find all of his work and regular commentary at bradlittlejohn.substack.com. He lives in Loudoun County, Virginia with his wife, Rachel, and four children.
NOTES
[1] Iris Murdoch, “Metaphysics and Ethics,” in Existentialists and Mystics: Writings on Philosophy and Literature, ed. Peter Conradi (Chatto & Windus, 1997), 75.
[2] Marshall McLuhan, Understanding Media: The Extensions of Man [1964], first MIT Press edition (Massachusetts Institute of Technology, 1994), 41–47.
[3] Anton Barba-Kay, A Web of Our Own Making: The Nature of Digital Formation (Cambridge University Press, 2023), 236.
[4] Barba-Kay, A Web of Our Own Making, 201.