Why AI Makes Sense of the World

In a world of armchair sociologists, ChatGPT fits right in

Have you noticed anything curious about the articles that are most likely to come across your radar in recent years? Consider the phenomenon of Aaron Renn’s “Three Worlds of Evangelicalism” essay, which appeared in First Things in February 2022 and enjoyed such a lively afterlife of literary discussion and cultural impact that it has an entire (quite lengthy) Wikipedia article dedicated to it, and went on to spawn a book, Life in the Negative World. So influential has Renn’s framework become that even its apparent refutation—the “vibe shift” following Trump’s re-election—has been discussed in terms drawn from Renn’s essay, as a reversion to “Neutral World.”

And yet the substance of Renn’s argument can be boiled down to the rather banal observation that America used to be culturally Christian, then went through a phase of postmodern relativism in which Christianity was one lifestyle choice among others, and sometime around 2010 began to turn actively against Christianity. We may wonder why such truisms merited “Book of the Year” recognition and NYTimes profiles.

But this is not to knock on Renn in particular. Many of the most talked-about essays of recent years seem to share a similar character. A few months before Renn’s essay, Mere Orthodoxy published a widely-discussed essay by Michael Graham, “The Six-Way Fracturing of Evangelicalism,” which provided a heuristic framework for plotting different wings of American evangelicalism along a cultural-political spectrum. James Wood’s hotly-debated 2022 First Things essay on Tim Keller, which spawned an extraordinary torrent of caricatures and counterpoints, is ultimately doing something similar, although framed as a first-person narrative: seeking to make sense of the changes in the contemporary evangelical cultural landscape.

As I was discussing this trend last week with a friend in a café in Vienna (yes, I know—*cue Napoleon Dynamite “Lucky” gif*), we realized something: we are all armchair sociologists now. Sociology is all about pattern-recognition in human behavior—trying to step back from the dizzying chaos of human experience, and discern some underlying order in it, as one steps back from the seemingly random dots of a pointillist painting to reveal a coherent picture. Rarely, I wager, have we been such suckers for sociology as we are today. We can’t get enough of it. The mega-hit book of last year, of course, was unabashedly a work of sociology, Jonathan Haidt’s The Anxious Generation, but even more theory-laden books of recent years on the Right have tended to take the form of “let me explain to you what’s going on around you right now”—which is to say, sociology: e.g., Patrick Deneen’s Why Liberalism Failed and Regime Change, Rusty Reno’s Return of the Strong Gods, Carl Trueman’s Rise and Triumph of the Modern Self, James Davison Hunter’s Democracy and Solidarity.

Again, this is not to knock on any of these books or essays—clearly they are scratching an itch, and many of them are deeply illuminating. But it’s worth noting what they do and don’t offer. While most of these works do contain some prescription, they get most of their traction merely from description: “I’m just here to tell you what’s happening.” After all, sociology is not in the business of investigating Truth, in the capital-T sense. Sociology tells you “here’s what people seem to believe these days,” not whether those beliefs are correct; it tells you “here’s what people are doing now,” not whether they should be. Aaron Renn’s “Three Worlds of Evangelicalism” essay could have easily been written by an atheist, rather than an evangelical Christian, and the basic observations largely unchanged.

Or consider the endless discourse around “Christian nationalism,” very little of which seems concerned with actually making concrete normative claims about the proper intersection of faith and politics. Most of it consists simply of spinning a sociological narrative about “the progressive gaze” or “white rage” that supposedly drives the political dynamics of our time. Moreover, in at least a lot of this kind of writing, what matters is not even the small-t truth of the factual details, which are often eminently contestable and hotly contested; what matters is the narrative’s ability to provide a sense of coherence or cognitive rest, to provide a comforting picture in the midst of confusing times.

Of course, in penning these reflections, I am engaged in a shamelessly ironic and self-referential (or perhaps, shall we say, “meta”) exercise. For what am I doing here but stepping back from the chaos and highlighting a larger pattern in the recent trend of our reading and writing? But I do have a larger point to make. For my friend and I stumbled on this observation in the course of a conversation about something else altogether: AI, and whether AI is genuinely a form of artificial intelligence. Your answer to this question depends in large part on what you think intelligence is. For the classical and Christian tradition, the uniquely human vocation of intellection is the contemplation of Truth, the union of the mind with the underlying principles of reality. To know a thing is not merely to observe it and form practical actions on the basis of it, but to know it as true. Can an AI do that?

It is hard to see how it could. What LLMs are so supremely good at, and what has enabled them to take the world by storm these past few years, is pattern recognition. They can take the almost infinitely vast range of data points with which we have filled the virtual world, connect dots, and distill them into an answer to almost any question or a solution to almost any problem. But is the answer true? Well, maybe yes, maybe no—that depends on the quality of the underlying data. Garbage in, garbage out, as they say. On many questions, one can trust them almost implicitly—if I’m trying to pull up birth rate statistics for the EU or a list of the highest mountains in the Himalayas. On other questions, such as, say, what caused the Great Depression, it’s not so simple. ChatGPT, faced with this question, will almost immediately spit out a well-organized, clearly-written set of bullet points under seven distinct headings: Stock Market Crash of 1929, Bank Failures, Reduction in Consumer Spending and Investment, Global Trade Collapse, Dust Bowl (for the US), Uneven Wealth Distribution, Poor Policy Response. A good answer, to be sure, and yet a rather noncommittal one! Almost certainly, all of these played some role…but then, which of these was the most important factor? To this, ChatGPT will respond, “Great question—and a tough one, because historians and economists still debate this. But if we had to pick the most important factor, many experts point to…” All the more so will ChatGPT dodge genuinely normative or metaphysical questions.

LLMs are consensus-seekers, not truth-seekers. They are, at least when it comes to questions of fact, essentially an artificial Wikipedia. And that is a very useful tool. In fact, it is an extremely useful tool in an age of absolute information overload. Since at least the advent of the printing press, among our most important technologies have been information management technologies, technologies which allow us to classify, sort through, and make sense of the increasing deluge of information that our technologies are generating. When, a few years in to the internet, many of us were beginning to drown in its sheer scale, Google became one of the most powerful companies on earth by creating a tool to bring order out of the chaos, to summarize the web for us. Two decades further on, the internet had ballooned to the point where this was no longer enough, and most of us felt powerless to navigate the world we had helped create. Along came AI, promising to tell us what the heck was going on, to simplify and summarize. When ChatGPT spits out its calm, cool, and evasive explanation of the Great Depression, it does so with the soothing voice of a therapist, helping us find meaning and transcend the anxieties of the Information Age.

We can’t resist AI, in other words, for the same reason that we can’t resist frameworks like “The Three Worlds of Evangelicalism.” They can’t tell us what is true, good, or beautiful, but they can give us some kind of cognitive rest by reducing a crippling chaos of confusing data points to something tidy, coherent, and eminently plausible. Our little minds were not made for such a world as this, and faced with its sheer complexity, we are forced to increasingly abandon the proper task of human intellection—that is, the investigation of what is true and how to live in light of it—and spend most of our time and effort in the preliminary task of mapping and remapping the landscape. Necessary though such a task might be, our grave danger is that we should rest prematurely in it, and mistake it for our true task.

Here, yet again, we encounter the feedback-loop phenomenon of modern digital technology: we create computers as a substitute for part of human thinking and acting, and then come to mistake that part for the whole, increasingly re-framing our own habits of thought in imitation of our machines. In a world where the best human writers no longer care to try and wrestle with questions of truth, and settle for crafting a coherent narrative, why should we be expected to notice the epistemic limitations of AI?

Recently Published

• “Family Formation and the Future” (the Danube Institute): If you’d like to watch the video of my remarks last week at the Danube Institute, along with those of my wonderful fellow panelists, and the fantastic ensuing discussion, you can find it here on YouTube. The whole conference was superb, and well worth your time!

• “Where Have All the Babies Gone?” (WORLD Opinions) If, on the other hand, you don’t have much time to spare, take just five minutes to read my reflections on some of the conference’s central themes, published yesterday at WORLD.

• “Called to Freedom” (Cross and Gavel): It was a real pleasure to sit down with Anton Sorkin of the Christian Legal Society to explore my new book on freedom in relation to questions of law and virtue—as well as talking about what seemingly every podcast host wants to talk about right now: technology.

Upcoming Speaking Gigs

• “Living Out of Control” (guest lecture at Patrick Henry College, 4/11): Today I’ll be making the short drive over to Patrick Henry College to talk about faith, career, finding a vocation in public policy, and…digital technology. Here’s hoping it all hangs together!

• “American Ideals or American Idols? Thinking Christianly About Freedom” (forthcoming event at the University of Northwestern, St. Paul, 4/15): I’ll be visiting the Twin Cities next week to give the inaugural lecture of their Classical Christian Honors College’s lecture series. Really looking forward to seeing many old friends there and making new ones.

• “Called to Freedom” (University Bookman, 4/22): I’ll be joining my friend Luke Sheahan, editor of the University Bookman, for a live webinar conversation about my book Called to Freedom: Retrieving Christian Liberty in an Age of License.

• “Called to Freedom” (St Peter’s Church, Purcellville, VA): Yep, one more talk about my book….Thankfully it’ll be a first for this audience, even if it’s the forty-first for me!

• “The Purpose-Driven Tech Life” (Academy of Philosophy and Letters Conference, College Park, MD, 6/6): When I’ve talked about tech issues over the last year, I’ve often gotten the question afterward of how our family tries to cultivate a healthy relationship to our devices. (“Tries” being the key word here!) I’ve often been surprised at just how surprised and intrigued people are by our household practices, and I’ll be using this speaking invitation to elaborate upon it not just as a practical how-to, but as a larger philosophy of technology.

Support my Work

Since writing is part of what I do for a living, and you, my readers, appreciate that writing, I’ve recently added a paid subscription tier for those who want to support my work with a small annual or monthly contribution. If you’d like to see more about how our technological culture is reshaping our understanding of what it means to be human, what we can do as citizens to fight back, and what we can do as Christians to chart a better path, consider a paid subscription. Most of the content will be available for free to all subscribers, but the full archive and occasional member-only posts (including “podcasts”/audio posts) will be for paid subscribers only.