One claim, four audiences, and the week it stopped needing a name
I · The Desk
Setting
The week began at a desk with a small task: an article for Vartija, a Finnish journal of religion and culture, about one chapter of Paul’s letter to the Romans. In that chapter a voice says “I” and describes wanting one thing and doing another. For centuries readers have argued about whose voice it is, and the argument has never really ended. I wanted to try something simpler than taking a side. I wanted to check the chapter against the story it seems to be retelling.
That story is the garden at the start of the Bible. With a language model I first wrote down what a verse would need to contain to match each stage of that story, and only then looked at the verses. The match was closer than I expected: Paul tells the garden story in the first person, using some of the same words. Nobody has to take my word for it, because anyone can put the two texts side by side. That was the point of the method.
What came out of it was a bigger claim than the chapter. The “I” in Paul’s letter is not only Paul, but anyone who has learned to speak and found that words change how we see ourselves. The journal accepted the article provisionally, for readers who know the old arguments well. It was meant to be one article among many, read mostly by specialists in the field. That was the starting point, and for a day it looked like the whole story.
Trigger
Then the journal asked the question every editor now has to ask: how much of this did a language model write? The article had taken only hours, but it rested on a reading of Genesis and Romans I had modelled many times before, in essays for A Tale of a New Era. The text was a research result. So I heard the question as asking how much of the result was the machine’s and how much was mine. A percentage of typing could not answer that.
The Greeks had a word for this kind of mistake: a shot that is aimed carefully and still misses. The question aimed at authorship, because that is how we have always trusted texts: we ask who wrote it, and the name stands behind the content. When a machine helps to write, the name stops meaning quite as much. The old measure was still in our hands, but it no longer fitted the thing we wanted to measure. It is simply the measure everyone reaches for first.
The reading itself pointed somewhere else, away from the author. If the “I” in the letter really is anyone who has learned to speak, that would be true whoever typed the sentence. If it is false, no famous name or title could rescue it. What mattered was not who wrote the article but whether anyone, with or without a theology degree, could check it. It took me the rest of the week to see where that simple thought led, and how far.
II · The Seat
Rising
The editor’s question stayed with me for a day, because it was fair but aimed at the wrong place. In the same weeks, AI systems were reported to have solved mathematical problems open for decades. Nobody asked what share of those proofs a machine had written, because a proof can be checked line by line. A machine misleads us mostly where we judge its answer while hoping for a particular result. There it behaves like a mirror that tells us what we want to hear.
So on Tuesday I sent an essay to the philosophy magazine niin & näin about the difference between plausible and certain. A language model is very good at the plausible: it predicts what most likely comes next. Something outside it has to decide whether the result actually holds. In mathematics that something is the proof; in reading the Bible there has usually been nothing like it. That is why a plausible reading of scripture can be so convincing and still be wrong.
The difference lies in when we agree on the rules. In mathematics we agree on the rules first, and after that nobody’s wishes change the answer: one plus one stays two. In interpretation we usually judge each reading on its own, so our hopes can creep into every verdict. Write the rules of a reading down first, let someone else apply them, and the question “who wrote this?” becomes “does it hold?” The editor’s question deserved a better answer than a percentage.
Climax
The turn came on Wednesday, somewhere between two drafts. If a checked result does not depend on who wrote it, then it also does not depend on who explains it. That sounds like a small step, but it changes the whole question at once. It was no longer about my article; it was about everyone who explains the Bible to others. For centuries some people have had the final say on what a passage means, and that position was suddenly in question.
So I wrote to the church itself. Kotimaa, the weekly paper read in Finnish parishes, received an essay with a question for a title: will artificial intelligence replace the church? Almost everything the church does, from baptism to confirmation class, rests on the Bible, and now a machine has read the Bible too. My guess was that the machine does not replace the church, only the chair from which its readings have been decided. Anyone can now sit in that chair and check.
This was the turning point of the week, and like most turning points it looks obvious only afterwards. At first the question was how much of a text belonged to me, the writer. Now it was who has the right to say what a text means. The answer is uncomfortable for any institution: no one in particular, because anyone can check. And yet, as I wrote to the parish readers, only a person can recognise a story from inside their own life.
Pathos
On Thursday the circle widened to Finland’s largest audience. Suomen Kuvalehti, a weekly news magazine, is the same magazine that in 1971 introduced Heikki Räisänen, a young Bible scholar whose findings upset the church. A long public debate followed, and one worry ran through it: what happens to ordinary churchgoers if scholars tell them directly what they have found? Many felt such findings should stay among experts. The line between the scholar’s desk and the church pew was the church’s way of protecting its people.
My essay asks what happens when that line no longer holds. The Lutheran church has a rule that the Bible alone decides what it teaches. So if anyone can check a reading of the garden story, and the text says something different from the teaching, the church’s own rule turns against the teaching. The church stopped treating the garden as history long ago, yet it still teaches that everyone has been born into a broken state “ever since”. Ever since what, if the moment never happened?
This was the hardest part of the week: the consequences no longer stay where they began. In the 1980s churchgoers could be shielded, because findings passed through experts who could be argued with. Now there is no expert in between, and nobody to shield, because anyone can compare a verse with a rule. After sending the essay I rewrote its opening overnight, starting from the magazine’s own role in that old debate. By then the claim no longer made a difference between a scholar and a churchgoer.
III · Four Reply Dates
Anagnorisis
Only on Friday morning, reading the four cover letters side by side, did I see what the week had been doing. The claim had not changed once: a reading checked against rules written down in advance does not belong to any one reader. What had changed was the audience, from scholars to philosophers, to parishes, to everyone. Each step removed one more layer between the text and the person holding it. The week had a plot, and I had been inside it without noticing.
I had already described the same plot in the magazine essay. Five hundred years ago, a new edition of the New Testament in its original Greek showed that a well-known verse had been translated in a way that suited church practice. Luther took that verse as the first point of his famous protest. At his trial he asked to be proved wrong from the Bible and plain reason, not by authority. Anyone who could read the original languages could check who was right.
Luther’s method gave the check to readers of the languages, modern Bible scholarship gave it to researchers, and the language model gives it to everyone. Each time, the gap between the text and the teaching became visible, and each time the church changed, even if it took generations. It is the same movement Paul describes: something is off, someone sees it, and things turn. I had traced it in the essays through the garden, Paul and the Reformation. On Friday morning I recognised it in my own week.
Denouement
The week ends with four cover letters and four reply dates. Three of the magazines have until the first of November to reply; Vartija has already said a provisional yes. Each will decide for its own readers, and some may well say no. That is as it should be, because an editor’s judgement is one more check, not an obstacle. Whatever they decide, the texts are already online, together with the conversations with the model that produced them, for anyone to read.
A week is a short time for a movement that once took generations. But what matters is not the length, it is the direction: from the desk to the pew, from the expert to anyone holding a phone. At every step the same reading reached a wider circle, and at no step did it need my name to hold. That is the part I find hardest to say and easiest to see. A text that can be checked by anyone carries itself.
In the garden story, the first question put to a human being is “Where are you?”, and the answer is to hide. Paul’s turn begins when a voice stops hiding and says where it stands. This week I tried to do the same with a reading, in four places and to four audiences. The outcome is not in my hands, but the checking is in everyone’s. Next week the replies will start to arrive, and the story will go on without asking who wrote it.
Written in dialogue with a language model. The four essays: “Mistä minä puhuu?” (Vartija, provisionally accepted); “Uskottava vai välttämätön” (niin & näin); “Syrjäyttääkö tekoäly kirkon?” (Kotimaa); “Mitä jos luenta on väärä?” (Suomen Kuvalehti). Sources: Heikki Räisänen, Paul and the Law (1983); Tuomo Mannermaa, Paralleeleja (1992); Suomen Kuvalehti 49/1971; Aristotle, Poetics; Erasmus, Novum Instrumentum (1516); Martin Luther, 95 Theses (1517) and the speech at Worms (1521).
The prompts, answers and corrections behind this entry, in English translation: The Measure of a Week — the dialogue behind the entry.
— Laurentius Paulus