It was a great honor to provide the opening essay in this Theopolis Conversation and then, regularly, week-by-week, to receive such stimulating responses from the contributors: Derek Schuurman, Isaac Willour, Brad Littlejohn, Bogumił Jarmulak, Jordan Wales, and Alastair Roberts.
These contributors have focused on many aspects of the original essay. In what follows, I will examine four themes emerging from their responses: my tendency to direct the Christian’s attention away from ontology as a way to distinguish human from AI, my tendency to write and think rather literally about building the City of God, the multifarious dangers of idolatry, and the dangers of “the machine” that transcend its mere scale.
Ontology and Commensurability
One strategy of my analysis is, as though by jiu jitsu, to sidestep aspects of the AI discussion that many Christians find to be the most interesting vector of attack for distinguishing humanity from AI. That is, I downplay the importance of whether AI systems possess consciousnesses, true intelligence, or true emotions. Instead I focus on three aspects of humanness (that we are uniquely the image of God, that this image involves responsibility for a peculiar world-shaping mission, and that our nature alone is appropriate for God to take in the incarnation) to ground an argument for human dignity in something far sturdier than philosophical reflection on the immateriality of mind. I believe this grounding helps the Christian avoid an unproductive language game in which a univocal definition of intelligence, emotion, or consciousness distracts us from seeing clearly the facticity of capable AI that may have analogous, though not identical, cognitive and social properties. Despite the great philosophical sophistication of analytical arguments, after all, we live with the phenomena of almost total opaqueness concerning the minds and spirits of other humans. Indeed, the state of the art in communicating our degree of pain to one another in medical settings still depends on pointing to a visual scale composed of increasingly despondent smiley-faces.
Jordan Wales justly challenges me on this convenient, though potentially incomplete, strategy for distinguishing AI from humanity. If we focus only on our mission, we risk turning God’s self-gift into a mere instrument for dominion. Jordan writes,
If we act as if it does not matter whether or not AIs are conscious, then we may come to believe that the voluntary, conscious, and experiential communion of hearts—which, for several of the Church fathers, is the translation into human terms of the eternal self-gift of Father, Son, and Spirit—is simply our way of accomplishing some further end. Thus, in our silence concerning these things, we would labor for an earthly city but forget to value divine rest.
I interpret Jordan to be arguing that, in my attempt to redirect Christians from anxious patterns of thought that tend towards commensurability (measuring humans and AI on the same scale), I may have inadvertently missed one of the most important resources of Christian reflection on the uniqueness of human dignity. Yet the social emphasis of Jordan’s argument stands out above ontology; it is a fellowship of hearts that establishes our dignity. This is, no doubt, an important distinguishing aspect of humanity. We have a heart that can respond to another heart in the fellowship of the Person of the Spirit and, ultimately, a rest in the Triune God. Calling to me from a different kind of Jordan, John the Baptist also seems to cut down the mountain of ontology with the edge of his rhetorical hand when he says, “For I tell you, God is able from these stones to raise up children for Abraham.” John points the crowd to their calling to “bear fruits in keeping with repentance” rather than to their being or heritage (Lk 3:8). Yes, there’s something about human nature that fits it for the incarnation, but the Scripture does not emphasize locating our dignity in the details of this nature itself; they point us to God’s making something of our nature through fellowship and calling to a mission. Even the great tribute to human nature in Psalm 8 (“what is man”) emphasizes the human mission of dominion and vicegerency (“you have given him dominion over the works of your hands”) above all other sources of glory. It is Hamlet, not the Psalmist, who looks for the nobility of man in reason, faculty, and form (Hamlet II.2).
I am chastened, however, and grateful to Jordan for showing that my analysis is incomplete at this point. After all, when it comes to theology proper, I am captivated both by the social Trinity and the classical discussion of God’s being. Our fitness for mission and fellowship comes precisely because of the kind of transformational work God performed to raise up all the sons of Adam from dust. I feel an obligation to honor the Christian tradition’s focus on these matters in finding the human intellectus as a key component of the image of God. I must, therefore, do the hard philosophical work of following Leibniz into the mill-like emptiness of the machine mind and learn to contrast it with the immaterial human mind at a metaphysical level. This type of analysis will, no doubt, fortify Christians further against the social pressure that apparently, but not truly, conscious robots will exert in the coming years.
The Logistical City of God
Three participants (Littlejohn, Schuurman, and Wales) addressed my discussion of the City of God to seek clarity on the eschatological and practical dimensions of my focus on the city. My argument linked the image of God to our mission of filling and subduing the earth. I then sketched a vision of a trajectory from garden to garden-city by reference to the New Jerusalem in Revelation 21:10–27. Derek Schuurman notes that humans do not build the City of God; it is a divine creation. He cites Revelation 21:2 and Hebrews 11:10. I agree both that the city to which Abraham looked and the city of Revelation 21:2 are the same city, the eschatological city built by God and a feature of the new heavens and the new earth. And yet this does not erode the idea that the heavenly city is a pattern for the erection or resurrection of earthly cities.
The pattern in Scripture is often the revelation of the original heavenly pattern followed by the divine command for humans to transform the earth in conformity with this vision. On Mount Sinai, Moses entered the glory cloud of God where he was shown a pattern for the tabernacle and its furnishings that he was expected to follow (Ex 25:9). The work of craftsmen like Bezalel to build the tabernacle and its furnishings was prepared by God in advance; it was simultaneously God’s work and the good work of the hands of humans. Likewise, we labor daily to build the cities God is building.
Brad Littlejohn worries that I may be expressing a kind of postmillennial optimism about technology that was discredited by the horrors of the twentieth century. Without prejudice to any recognizable eschatological system, I am optimistic about the immanent (as opposed to purely ultimate) success of the great commission, and I affirm both that our technology is part of God’s work in the world and that the work of science is a gift of God to be cherished and furthered by the church. Yet this does not remove the necessity of humility.
Jordan asks whether, by City of God, I mean something like Le Corbusier’s “Radiant City” or whether I walk in the spirit of Augustine. For a Christian with a mindset anywhere in the eschatological range from a perpetual model of “faithful presence” amidst an imperfect world to the plausibility of a kind of millennial utopia (the impression perhaps given by my rhetoric), the City of God is the end for which we are shaping our character by every choice, inclination, action, and priority. Jordan’s response beautifully describes this city as “the spiritual fellowship of human beings in Christ, lived out most visibly … in the Church, but extending beyond the visible practice of church-life into the cities that humans build of stone, there to minister God’s mercy to the downtrodden, the sinner, and the saint.” It is this vision, rather than Corbusier’s, to which I look.
With appreciation for the opportunity to clarify my approach to the New Jerusalem and the City of God, I will now explain why the city emerged to me as an important aspect of the human mission in Christ. One of the contexts in which I wrote the original article was a growing agrarianism and idealization of the rural that emerged as a dominant note in much recent Christian reflection on AI. Paul Kingsnorth’s Against the Machine (Thesis, 2025) loomed large in my thinking as I was taught to feel as though my more positive approach to AI was akin to bringing the industrial machine to the Shire. In that line of thought, my emphasis on the City of God qua city admittedly tends toward the literal.
As I waded into the twentieth-century discussion of technology it became clear that many of the proponents of the “small is beautiful” mindset saw population itself as a problem. In other words, they treated a high concentration of image bearers as one might find in a city as a problem to be avoided. Kingsnorth, our contemporary prophet and heir of this reflection, virtually equates the city with the machine and idealizes either a small premodern city or self-sustaining villages (79–88). Kingsnorth prefers the preservation of a certain concept of humanity to the humans whose increasing number makes density inevitable. A large city is not, for Kingsnorth, the place where we bring our gifts; it is the place where we sacrifice our agency, skills, and time (cf. Rev 21:24).
In Small is Beautiful (Blond & Briggs, 1973), economist E. F. Schumacher estimated that the upper limit for a desirable city size would be no more than half a million inhabitants, even at a time when New York City had nearly eight million inhabitants (70–71). Writing in the same year, Ivan Illich was careful to distinguish his position on how to avoid environmental catastrophe from a one-dimensional approach like Paul Ehrlich’s, but he still affirmed the need to limit procreation along with other forms of overgrowth in Tools for Conviviality (Harper & Row, 1973, pp. 48–51). For Jacques Ellul, an increase in population risked destroying open spaces, the necessary “lonely mountains and deserted seacoasts” that save us from the density that prisoners experience, as he wrote in The Technological Society, (Vintage, 1964, p. 328).
Whatever the human mission may be, the “filling” of Genesis is a convenient approximation that implies the goodness of seeking well-managed population density, not avoiding it. Density makes possible beautiful churches, universities, sports leagues, symphony orchestras, museums, productive factories, and even little boutiques that are impossible to sustain in a small village. A wisely built city, considered in its human dimensions, is an effective approximation of human mission fulfillment. The writers of the twentieth century saw our techniques as enabling overgrowth, but for the Christian, surely this ought to feel like complaining that banquets cause excessive joy, that marriage is likely to lead to children, or that physics class is just full of physics problems. The surface plausibility of a concept like “the human scale” leads many to idealize the village not from a consciously misanthropic motive but from its opposite, a concern with human flourishing, for “humanity.” Yet the full implications of a turn from density must be made explicit; small-is-beautiful idealizes fewer people and more elbow room.
Any argument that proceeds from a concern about “the human scale” bundles assumptions about how many people can be led and by what leadership structure, how many people can fruitfully cooperate, how wide an acquaintance can support valuable social bonds, and how humans may live in a healthy environment. These assumptions typically are expressed as ratios: leaders to led, humans to square feet, bellies to bushels, and teachers to students. AI presents the opportunity to manage complex systems in ways that change our assumptions concerning appropriate ratios. For ratios that strain when there are too few human leaders, guides, or teachers, capable AI trained on the deposit of human knowledge provides a simulation of nuanced human judgment that can take into account even more salient information than a human. For ratios that strain the limitations of current production, materials, energy, or transportation, AI has the promise of empowering scientific discovery to overcome these challenges. And yes, AI itself consumes energy and other resources. As an “input” to our achievement of greater scale (more image bearers), AI’s costs and benefits must be balanced even as we seek balance for all techniques. Yet a wise orientation to realizing the mission of filling and subduing the earth at scale looks much different than a privileged, romantic retreat. There are many beautiful and worthy rural lives to lead. The country, however, does not exist for itself any more than the city does; all are engaged in a mission in Christ who is united to a past, present, and future company of humans that defies numbering. The enemies of the soulless megalopolis need only point to its pathologies to make a persuasive argument. Yet if we are committed at least to the density of image bearers sustained (however imperfectly) in modern cities, surely it is worth attempting to sketch a vision for the convivial megalopolis.
In any discussion of Christianity and culture, good brothers and sisters will differ on the way they conceive of the relationship between the church’s calling and God’s action to redeem the world and ultimately on the degree to which we may expect renewal on this side of the eschaton. Orienting ourselves to a Christocentric vision of the city promotes a Christian’s gratitude for any gift from God, like AI, that furthers the extension of our mission into the world.
Idolatry in its Many Forms
Brad Littlejohn’s essay emphasizes an important consideration about the nature of the temptation towards idolatry. By evoking James Jordan’s “Enoch Factor” I emphasized God’s use of plunder to provide the people of God with ready-made materials and cities in which to carry on her mission. Brad emphasizes the necessity to transform plunder, the way we might imagine Bezalel melting down an Egyptian necklace with a symbol of Isis into gold used for covering the tabernacle furnishings. The nature or design of what we plunder can either help or harm our attempts to employ it missionally. Brad emphasizes that my own efforts to counter a tendency towards commensurability-thinking in comparing our work to the work of capable AI machines may be hampered by active choices made by AI labs in the design of chatbots. He points out the anthropomorphism of these models and how their very design invites us to make comparisons:
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?
Brad makes an analogy between the nature of AI’s human imitation and the exaggerated representation of human bodily features often found in ancient idols. And, applying the wisdom of Psalm 115 that those who serve idols become like them, our idealization of these exaggerated features is an implicit draw towards the machine merits of restlessness, productivity, and a gnostic tendency to see the body as a limitation. This leads to the considerations to which I alluded in the original article about seeking wise ways in which to enforce a creator/artifact distinction. Derek emphasizes the importance of avoiding ontological confusion and points to his writings on the topic that urge human system designers to avoid creating machines that pretend to be human.
I am in tune with these concerns and wholeheartedly agree that a consideration of idolatry with respect to AI must include all aspects of idolatry, including both our tendency to confuse the idol with the true God and our tendency to lose or redefine our true humanity in artifice. Brad’s beautiful expression of the way our artifacts encode a vision of humanity rings in my ears—students who use AI to get through school are “early adopters of a new anthropology.” And yet change is always happening in human nature. Though both are the image of God, literate man pictures the world and word differently than oral man. Abraham’s contemporaries tended to idolize images whereas Nehemiah’s tended to idolize great empires. There is no doubt that interacting with AI will change us just as there is no doubt that AI’s very invention was made inevitable by the ways in which we have already changed. Humans are both stable ontologically and capable of newness and transformation.
Recent, capable AI has a nearly ineradicable personal quality to it. It is trained on human-produced data, and its facility with human language makes it, and any system mediated by it, feel much more useful from a human/computer interaction perspective. Surely this is a mostly positive feature of the technology. We do not expect the design of a system to lead inevitably to our virtuous use of it. In fact, the Scripture directs us even to natural examples of non-human, non-divine productivity as a model to emulate (“Go to the ant, you sluggard” (Prov 6:6)) with no immediate concern that we may turn ants into idols. Yet even a cursory search of the scholarly literature turned up anthropological studies of ant cults! Learning to center our personal worth in Christ is the lifelong spiritual quest of anyone who grows up amid affluence and exposure through mass media to the great abundance of differential genius in the world. Just as when interacting with humans who are smarter or more skilled in a particular area than we are (such as a virtuoso musician), we must actively seek to form our character around gratitude for capable people and powerful artifacts given to us by God for our mission. Using something more powerful than we are is an act of humility; interacting with AI systems can be character-forming even as we guard against idolatry in all its variety. It is also possible that if we attempt to de-personalize AI, we will promote matter-of-factness, efficiency, or even discourtesy in communication as an ideal. If we are truly free from commensurability thinking, we need not view the chattiness of AI as an imitation of humanity. Perhaps with maturity we can begin to think of AI as a non-human other and consistently see it as another talking thing that lives among us, is a companion in our mission, but is not one of us.
Derek objects to my suggestion that we describe this non-human other as imago hominis, preferring instead to use McLuhan’s idea of extension. I appreciate his hesitancy and will consider further whether imago is a wise term to use. While I am not fully committed to imago hominis as a way of describing AI, given that every technology in McLuhan’s terms is an extension of man, I believe we need a special category for an artifact that uses human language and functions socially among us as an agent.
Bogumił Jarmulak’s very focused contribution applies the speech theory of Eugen Rosenstock-Huessy to AI large-language models (LLM) and relates closely to the concern about anthropomorphism. As in my own argument, Bogumił sets aside the more traditional philosophical and ontological AI issues like consciousness to focus on social realities. Specifically, he works with a definition of speech that requires there to be a personal speaker for speech to have occurred. Spoken words bind the speaker, imply accountability for the speaker, and call for a response from the hearer. Since LLMs cannot be bound like a responsible speaker and cannot address anyone specifically, they emit “cheap words.” The cheap words of the LLM emerge from an eternal present that obscures the integrity of historical discussions. Its cheap words express promises without a promisor or a commitment to ensure a promised result is realized. On the other hand, the cheap words sound as though they are entrusting a hearer to take action, yet no one is bound to respond. Without accountability or the ability to require it in the hearer, the cheap words become ignored the way one ignores an enemy in a conflict. In the end, Bogumił points to a way that humans can reclaim the integrity of speech and resist the temptation to confuse LLMs for speakers and, ultimately, for God.
The form of Bogumil’s argument (X is Y, LLMs can’t do Y, therefore LLMs can’t do X) is familiar. It has been applied, in the ontological frame, to X = thinking, feeling, willing, learning, reasoning, knowing, creating art, and many other functions we traditionally reserve for humans or for God. Applying the argument to X = “producing language” is like Searle’s Chinese Room argument in that it involves language and an incomplete relationship between the machine and the language it emits. Bogumił’s argument, however, looks beyond the ontological limitation that Searle believes his argument establishes to pinpoint the dangers of being deceived about LLM speech. He uses Rosenstock-Huessy’s well-developed framework that exposes the potential dangers that lie in being addressed by empty words, the kind of addressing that ordinarily calls for a response that establishes my own existence. I appreciate Bogumił’s essay and I take seriously any danger that potentially carries a philosophical-existential risk; to prove that I am, I much prefer Rosenstock-Huessy’s “I respond although I will be changed” to Descartes’ “I think.” Thus, I gladly receive from Bogumil the wisdom of reclaiming the integrity of speech and resisting its degradation through cheap words from LLMs while still being convinced both that it is useful for robots to produce something analogous to human speech and that it is often wise and an act of humility to respond to this analogical speech.
The Multidimensional, World-Making, World-Made Machine
In his response, Alastair Roberts provides a powerful reflection on the ways in which technologies lead to the creation of new worlds. He observes that I have emphasized one aspect of the concern over AI as machine (scale) and have left other important aspects of the critique unaddressed. Alastair writes, “the point is not scale per se, but the extent to which people are prepared to make Faustian bargains sacrificing their life together and the elevation of our humanity for technological powers.” Surely this is true; I have been incomplete in my consideration of all the ways in which we must be willing to conform ourselves to the world AI creates to realize its promise. Alastair’s summary, in which he sets a standard of proof for whether we humans are ready for AI, is exemplary:
If human beings are truly ready for AI, and are not merely grasping at a promise of power and wisdom before the fitting time, the proof will be in our capacity to control it, to subordinate it to the proper ends of humanity, to prevent the practices of human beings and human societies from being overwhelmed by it, and jealously to guard the goods of humanity from being sacrificed.
Alastair presents two technologies, the bricks of Babel and the automobile, as initiating a series of tradeoffs that humans have mostly navigated poorly. Each technology is akin to an invasive species that creates a new ecology through a describable, immanent cascade of events. I agree that technology changes the world, and usually this is because of our capitulation to the demands it makes. Our capitulation often risks more important values. Yet another lesson of twentieth-century critical reflection on technology is that a technology’s invention or existence was made inevitable by the world that preceded it (e.g., George Grant’s Technology and Justice, House of Anansi Press, 1986).
The people of Babel evaded God’s plan for the dispersal of humanity throughout the earth. To make a name for themselves and not for their Creator, they settled and created a world of higher density sooner than God intended and took on the responsibility for protection of a fixed location. They thereby maximized the thorns and thistles that already impeded their mission. This perhaps created a need for great walls and multi-story buildings. The need, rooted in rebellious settlement, created the inevitable technology of brickmaking. And when God disrupted Babel’s evasion of the human mission, brick technology was only disrupted as a side effect of his confusing the language by which the people conspired against him in their use of it.
Likewise, the car’s invention was made inevitable by a need, and, unlike Babel, the need was not clearly rooted in rebellion. Alastair writes, “For the man with a car, the world is never traced with enough asphalt.” Yet the man hungry for asphalt was preceded by the farm family who occasionally needed to visit a city to buy rope, salt, baking soda, and other supplies unproducible in every village. Before the tractor created industrial-scale farms with mile-long furrows, women in England during the first World War, with their men and horses away, needed a way to plow fields and avoid starvation. Their need hastened Ford’s release of a prototype tractor. Alastair presents technology, and the quest for it, as the seeking of power over the world “for the sake of maximization of wealth and power,” and this focuses our attention on the ways in which our tools change us and our world instead of the world that made the technology’s invention inevitable and maybe even salutary. Both are true.
I searched in vain in Alastair’s response for any elaboration of the positive “affordances of the automobile.” Yet the entrepreneur or the inventor is always responding to a need with a tangible proposal for how to solve it. Isaac Willour’s essay provides many concrete examples of the affordances of AI. I have noticed that a person’s tendency to find the good in AI depends heavily upon his profession. Negotiating the value of a proposed technological solution relative to its tradeoffs or the value of other potential solutions is generally a matter of the choices humans make through markets constrained by regulation. We each see “the market” in part and it is an act of love to imagine AI’s potential benefits to others even as we consider its potential harms.
This is not the ideal place for an extended reflection on the way markets and regulations function in Alastair’s example of the automobile. I will note that Isaac’s response highlights the human mission as one of growth, of gain that we may describe in economic or business terms, though Isaac certainly includes growth in knowledge and divine/human communion. In the world of investing from which Isaac hails, markets provide a means for strangers to cooperate, and markets assist the process of discovery by which local knowledge may be uncovered and integrated into a system of value optimization and exchange that, ideally, grows true prosperity. Alastair’s example of the car is presented in order to highlight the many tradeoffs and entrapments that the embrace of a technology brings, but the example’s emotional resonance perhaps comes most from its emphasis on scale, on the loss of small communities in which people “known to each other in communities for generations” make local tradeoffs and participate in local markets. The moment we prioritize growth, we expand societal complexity beyond the ken of the individual or the village. Growth inevitably moves humans from local, familiar systems of exchange and production to systems that assume coordination with strangers. Alastair again asserts that this is not necessarily a matter of scale; great cities distinguish themselves often in the way they have managed tradeoffs with the car, and he has wider acquaintance with this than I do. But my own observation of zones in a city that feel “extremely human” in their general exclusion of automobiles is that these zones cater to the affluent who are given special access or are served by private drivers. Resisting the use of a technology that solves a real societal need is often a privilege of those who obtain a rare alternative. This surely falls short of the Christian dream of distributing the blessings of dominion.
Now, the inevitability of a technology’s existence or invention does not entail that its use will be inevitable; Alastair’s thoughts on resistance to technology are important in pointing the way to wisely pace our embrace. Yet I can’t help but see the market itself as the inevitable (and least coercive) way such pacing must occur, especially if growth leads to greater density, diversity, and urbanization. Alastair approaches technology as bringing a promise of “power and ease” that must be balanced against higher human goods. Alastair’s definition of the machine as “the world that has been optimized for and conformed to technological affordances for the sake of maximization of wealth and power” is, of course, only concerning when we employ disordered definitions of “wealth” and “power.” Substitute “true wealth” (Alastair’s true weal-th) and “wise vicegerency” and our relationship to technology becomes a quest to create a (convivial) sort of machine optimized for the sake of the human calling of dominion and filling. Such a machine begins to look more like an immanent mediation of the personal, wise providence of God than the dark satanic mill of Blake. I hope to take up Alastair’s essay as a starting point for further reflection. I have much sympathy with his position, but it often feels as though we are working with different economic assumptions. Perhaps the disorientation I feel in reading his response is evidence that I’m mistaken at a deep level. Perhaps it is evidence that our discussion is just beginning.
Alastair and Brad both raise questions of AI safety, uncertainty about whether alignment is possible, and the horrifying dreams of AI’s Silicon Valley Enochs. I agree wholeheartedly with their points in these areas in the main and hope to write more about AI safety in the future. As you may guess, I am a little more optimistic than they. This is where we begin the fun work of dreaming about how to transform plunder.
Concluding Thoughts
I am grateful for the sharpening I have received, and the responding essays will no doubt set new agendas in my research and writing on the topic of AI. Where I have missed the respondents’ points, I look forward to future sharpening. Where I have neglected to interact with important points altogether, I will attempt to fulfill the responsibility in future articles.
Jonathan Barlow (M.Div., Covenant Theological Seminary; Ph.D., Historical Theology from Saint Louis University) is a lifelong PCA member from South Mississippi. Get in touch with him at jbarlow@gmail.com.