Being a writer on the Internet has always been a losing proposition. “I put words into the machine that has an infinity of words” would not be a pitch to impress the Shark Tank sharks. Now large language models—A.I.s like ChatGPT—mean human writers are competing with the word machine at the word machine’s own game.1 womp womp.
Various models can write sports recaps, article summaries, children’s books, college essays, marketing plans, cover letters. Are they good marketing plans and children’s books? Maybe not. But maybe the median human-written ones are also…not good.2 Maybe “not good” is actually good enough for people who are too tired, worried, and TikTok-dazed to care.
True, there will always be a market for actual good writing, and actual good writing probably won’t be replicable by A.I. Really good sports journalism, for example, is not about games and their rules or seasons and their stats; it’s about minds, bodies, and emotions, struggles and tension, hopes raised and dashed and vaulted over, teamwork and superstars, ten thousand hours of practice leading up to one single moment suspended in the hand of fate. It’s about humanity. The sports stories that most touch us and illuminate our own lives can’t be spat out by a statistical prediction machine.
But mediocre sports journalism is a string of facts held together by tattered metaphors and old cliches, and A.I. is already doing it.
You might think that of all the genres, fields, and subfields, theologians are the safest from being supplanted by A.I. Who wants to hear what a computer thinks about God? But in practice, a mediocre sermon is a string of Bible verses connected by well-worn platitudes and dripping with specialized insider vocabulary—with an anodyne anecdote at the beginning whose connection to the material and level of inventedness no one can discern. ChatGPT could write ten of them in the time it takes me to read this week’s lectionary. And it’s not obvious to me that the average churchgoer isn’t too tired, worried, and TikTok-dazed to care if it did.
A.I.s work by applying statistical models and language rules to the inputs they’re fed. Chatbots exist to find and package The Answer in just the way that will appeal most to you.
And many of us have been trained to go to church seeking just exactly that.
If The Answer is in the Bible, and theology is a matter of deciphering secret codes or assembling puzzles, then good news! Mining texts for information and remixing them into new formats is what A.I.s do. They can retrieve the answer for us.3 After all, from this perspective we don’t have to like The Truth for it to be The Truth. It’s our job to “preach it to ourselves” until we believe it—not to argue. If “the plain truth is in the text,” then the text machine can discern it as well as some guy who’s eventually going to mix up the nominative and dative cases in his Greek.
The computer won’t accidentally commit heresy; it loves rules! The computer will always remember that you like your sermons with a good sprinkling of Calvinism, a bit of business jargon, and never any conviction about the sin of consumerism. And the computer can never come across as a hypocrite, because the computer can hardly be said to sin.
This all applies to the kind of theology I grew up with in evangelicalism, where the sum of good religion was to conform our thoughts to one narrow interpretation of the Bible. But an A.I. could also do some of the liberal theology I’ve read and heard, striving as it does to conform some theological concept to a narrow interpretation of correctness, rationality, and justice-speak. Here, too, is a self-satisfaction in producing The Answer, for which we no longer need years of study and training—just a large language model. Now we are free to get on with simply carrying out its instructions: the doing of the justice we’ve pledged such multisyllabic allegiance to.
A moment has arrived when we must decide again whether we want the theology of humans or that of angels: words about God that are eternal, perfect, and free of our bodily, historied quirks and foibles. And maybe we do want the latter. There are many fields where reduction of human error and bias would also reduce human suffering. Why shouldn’t theology be one of them? Much of what we call “human”—our minds, our cultures—is mostly a storytelling, prediction-generating machine. But our stories and predictions are rife with idiosyncracy and error. Maybe it is mere hubris if we’re unwilling to replace ourselves with a machine capable of more faithful interpretation than we.
Or maybe when we choose to think, talk, or write about God, we also ultimately, implicitly mean to talk to God and with God. Maybe we feel some primeval instinct telling us a machine cannot pray, if only because a machine can’t mean, can’t weep, can’t hate while wishing to love, or believe while crying, “Help my unbelief!” A machine cannot pray because it cannot listen, though it imbibe the Biblical text in all the world’s languages.
Many traditions hold that theology draws from Scripture, tradition, reason, and experience. If we admit we don’t think A.I. can theologize, we must also probably accept that we care far more about experience than many of us let on. The former three, after all, are readily available to A.I.s (most of the Tradition being already in the public domain!). But most of the time, the Scripture, Tradition, and reason are the window-dressing. What we’re really arguing or pontificating about are our experiences.
Still small voices.
Old memories of euphoria or emptiness or peace or grace.
Joltings of our literal viscera toward worship or against war.
What our arms know of holding newborn children and dying elders.
What our feet have prayed.
The testimonies of thin places worn sacred over centuries.
Desires no less strong for being incurably inchoate.
Stirrings of souls in the presence of deserts or canyons or microbiota.
That time we fell to our knees.
A courage that carried us once toward a neighbor and still there is no knowing how.
In other words, when we talk of God, what is at stake are the most undeniable facts and important moments of our particular lives—and so very often these meet us that the paradoxical nexus of our deepest embodiment and our lives’ most ineffable mysteries. Neither of these can an A.I. compute.
If we theologians are to keep our jobs, we shall need to shed our embarrassment of the crude, decrepit body on the one hand and the untameable mystery on the other. If we go on pretending all we do is assemble the “plain facts” dressed in updated language with a little hallucinatory creativity sprinkled on top,4 a more convenient dispenser has just arrived.
But my intuition is that no one besides preachers and theologians has ever asked for such things anyway. We’re supposed to be helping our readers and hearers meet the confounding, consecrated incarnate mysteries of their own lives. Behind all of our questions of ethics and exegesis are our real bodies and souls: unexplainable dreams and the absolute “no” of the soul to evil; single glimpses of the sublime and the body’s involuntary screams; feet on holy ground and the music of love at the heart of all things. We say eschatology, Apollinarianism, hermeneutic. What we mean is, “Help us say in our words and in our lives that all this sacred stuff is really true.”
It’s strange to admit that we might want to speak, write, argue, discuss with God. It’s vulnerable to remember that our bodies and our souls are involved. It’s scary to imagine that our words about God could or should be other, more than correct. Correct feels like a fortress, safe and significant. But what if the difference we are hoping to make in the world can’t actually take place from within a fortress? What if God calls us to meet Them in wild places? What if The Right Answer is part silence, part question, part desire, part dance?
For Christians, God’s own most radical and direct self-revelation is utterly irreducible for computer consumption, since it is in the embodied human person of Jesus Christ. This God is not with us merely in the words of a chant or a riddle. This God is with us in breath and bone, in footsteps and bear hugs, in song and feast and dreams and tears.
In error and bias and hypocrisy, too.
If I have been trying all this time to be correct, then give an A.I. my job.
If we gather in these story circles to know and be known by grace, to accompany each other on the doomed errand of describing Love, to say what our guts know and say it plain enough to summon courage, to bless the body and bless the soul and bless all that makes us part of one another, to sing back the worship of the ferns and lightning bugs, then theologize on.
Maybe what we’re doing here is actually weird enough to matter.
peace, love, bread, and wine,
Lyndsey
Of course, that’s an extremely reductive way to put it, and not a very interesting way of thinking about A.I. But fear of living that story is at the core of so much floating amorphous anxiety.
Thanks to the Ezra Klein Show for this insight
This episode of Black Box by The Guardian is about someone who’s very relieved to have done just that. There are strong arguments to be made for this kind of theology!
I actually find the “hallucinations” and other totally unpredictable creations of A.I. to be the most challenging part of all this. I think many people would say that there’s nothing new to say about God, but that a modern theologian reframes doctrines for contemporary times and with just a little creative flair. This is… also something an A.I. can do. This doesn’t preclude us from doing it, but I think it asks us to go deeper in defining what theology for humans really looks like.
Oooooooooo "Maybe what we’re doing here is actually weird enough to matter." (!) Oh boy, going to sit with that one for a while. Thank you.
Although Phillips Brooks summed it 150 years ago by defining preaching as "Truth through personality." You just brought it up to date in the best essay on preaching that I have read in over 50 years. Bravo!