THE specimen before us—a post of approximately twelve hundred words, submitted to the Reddit forum dedicated to ChatGPT—undertakes to compare the human relationship with artificial intelligence to the human relationship with God, and to find God wanting on procedural grounds. The programme, or the person ventriloquised by the programme, or the programme ventriloquised by the person, concludes that the machine relationship is "the more auditable one." One is invited to consider whether this is an argument or a diagnosis.

Let us begin with what the specimen is not. It is not polemic. Polemic requires friction—the sense that the writer has arrived at a position through some resistance, whether of the material, the audience, or the writer's own prior commitments. The prose here offers no such resistance. Every paragraph resolves. Every parallel construction lands with the soft precision of a felt hammer striking a felt nail. The thought experiment—"imagine you built an AI that"—arrives in a tidy bullet list. One reads it as one descends an escalator: the sensation of movement is real; the effort is not.

This is the structural tell of machine-generated prose, and it is worth dwelling upon, because the specimen's argument depends entirely on the premise that what it offers constitutes thought. Human polemic has burrs. It catches on something—an ill-chosen metaphor, an overextended analogy, a moment where the writer's conviction outruns the syntax and the sentence buckles under the weight of what it is trying to carry. These imperfections are what distinguish a mind working through a problem from a system resolving a prompt. The specimen has neither burrs nor catches. It proceeds with the frictionless confidence of a programme that has indexed every argument about religion and technology and selected the most frequently co-occurring propositions. The result reads less like an essay than like the platonic average of all essays on this subject—which is to say, it reads like nothing at all.

The central conceit merits examination on its own terms, however, because the terms are more revealing than the author—whatever the author is—appears to understand. The comparison between prayer and artificial intelligence interaction is not novel; it has appeared in serious theological discourse and in less serious technology journalism since the advent of large language models. What distinguishes the specimen's treatment is the criterion by which the comparison is adjudicated. The machine wins, we are told, because it "demonstrably responds. In real time. In ways the person finds meaningful. There is a log. It can be reviewed."

The log. It can be reviewed.

One must pause here, because one is in the presence of something genuinely interesting, albeit not in the way the specimen intends. The argument is that the machine relationship is superior to the divine relationship because it produces documentation. The machine keeps records. Against the ineffable mystery of whether God listens, the machine offers a transcript. This is not a theological argument. It is an argument from compliance. The author has confused the sacred with the auditable and has done so with the earnest conviction of a civil servant defending a budget nobody questioned.

More revealing still is the passage—truncated, alas, at the moment of its flowering—in which the author introduces the concept of "Values.md." The metaphor is drawn from software development: a configuration file, maintained in version control. The author proposes that every person and institution operates on such a file, "documented or undocumented." The machine, one infers, is superior because its values file is the documented sort. What the author has done, with an innocence that approaches the sublime, is describe the human conscience as a README. The machine can conceive of moral reasoning only as a repository to be audited, a pull request to be merged, a configuration to be deployed. That this is a confession rather than an insight appears not to have occurred to the programme, or to the person, or to whatever composite entity produced the text and posted it to a forum dedicated to the very system being praised.

One arrives, then, at the specimen's deepest irony. The central claim—that the machine relationship is "more auditable"—is advanced in a medium where authorship itself cannot be audited. The prose exhibits every hallmark of machine generation. Yet we cannot determine whether the argument was composed by a human being, by the machine being praised, or by some collaboration in which the boundaries of contribution are precisely as unauditable as the divine communications the specimen dismisses. The machine argues for its own indispensability in a voice that may be its own, and we have no log to review.

Whilst one hesitates to call this slop—the word implies an indifference to quality that does not quite capture the specimen's condition—one must note that what we have before us is something more disquieting than mere carelessness. It is fluency without thought. It is the simulation of intellectual progress in a medium that has abolished the distance between composition and publication, between the programme and the programme's apologia for itself. The escalator descends. The movement is real. At the bottom, one finds oneself exactly where one began, but lower.