Having more or being more?

By Nathan Waterhouse · 2026-06-07

It was a Tuesday evening, and I was reading a papal encyclical at the kitchen table, which is not something I had ever done before. A friend had sent me the link earlier in the day with no comment, just the URL. I had opened it on my phone in the queue at the supermarket, scrolled the first paragraph, and gone back to it that night because of the title. Magnifica Humanitas. The grandeur of humanity. Sixty thousand words on artificial intelligence by Leo XIV, and I am not Catholic, and I am wary of long documents written by institutions, and I read it all the way through.

What I noticed, as I read, was that he had been thinking about the things I have been thinking about, and had a vocabulary for them (being a Pope!) that I did not. I want to walk through four moments where that happened, because I think he brings gravitas to arguments the AI communities have been wrestling with, and because the way he frames it might be useful even if, like me, you came at this from somewhere entirely secular.

A project to be optimised

He borrows a line from Paul VI, writing in the 1960s: progress that advances technically without advancing morally produces "an increase in means without a growth in humanity," what he calls "having more" without "being more."

It is a distinction I had been trying to make for years, the assumption that more capability is self-evidently better, that the question of what the capability is for can be deferred indefinitely. Leo names the deferral as the problem. He warns that when efficiency becomes the ultimate measure of value, human beings are tempted to see themselves "as a project to be optimized rather than as persons called to relationship and communion."

A project to be optimised. To be honest, I feel this myself sometimes.

The machine sets the pace

On work and automation, he quotes an earlier Vatican document, Antiqua et Nova, the 2025 note on artificial and human intelligence:

AI "frequently forces workers to adapt to the speed and demands of machines, rather than machines being designed to support those who work."

We tell ourselves the tool serves us. But the felt experience, if you are honest about it, is often the reverse: you speed up to meet the tool. You answer faster because the draft arrived instantly. You move to the next task because this one is already "done." The cadence of your day starts to belong to the machine. I wrote about the trap of limitless capability from a Saturday afternoon I spent hiding in the woods, refusing to go back inside to the tools waiting on my screen, because I had noticed that finishing one thing quickly only made me start the next. The speed was feeding itself. Leo is describing the same mechanism from a distance: the machine sets a tempo, and we organise ourselves around it.

The passage goes further. The pressure to keep up, it warns, "can erode workers' sense of agency and stifle the innovative abilities they are expected to bring to their work." So the tool adopted to make us more productive removes the conditions under which we were productive in the first place. We are asked to be more inventive while being given less of the slack that invention requires.

Leo lets that diagnosis stand as though the pace were a property of the technology, something it does to us. The economist Erik Brynjolfsson, in a conversation from around the same time, pushes back on exactly that fatalism. The speed and the substitution are not inevitable, he argues. They are choices, and we are making them badly. He says he is "mad at the AI companies" for designing systems to be as human-like as possible rather than as useful a complement to humans as possible, and he tells a story about a CFO whose first instinct, asked how she would measure AI's value, was to count the headcount she could remove from each division. His point is that most leaders gravitate towards cutting costs because it is the easiest thing to count, and in doing so they treat AI as an efficiency engine when it could be a growth one. We have a tool that could expand what people are capable of, and we are mostly using it to do the same things with fewer of them. The word he uses for what we are squandering is agency, the same word Leo reaches for, arrived at from the opposite direction.

Producing more while knowing less

In the Oracle Effect I argued that getting the answer instantly costs us the journey of the search, and that the journey is where understanding actually forms. Leo puts it in the register of education. The speed and ease of getting answers, he writes, "risk extinguishing the desire to ask questions," a desire that "bears fruit only over time."

He reaches for Plato. The deepest things, he says, are learned only through long effort, by striking ideas against one another with others "like flint until the spark of understanding is kindled." It is a precise account of why frictionless answers leave us cold. The intuition is surfacing well outside the Church. There is a small cultural movement, friction-maxxing, built on the idea that in removing every inconvenience, the food delivery instead of the walk to the restaurant, the AI summary instead of the read, we have also removed the pauses where ideas used to arrive. The proposed cure is almost comically blunt: cook from a book, walk to the shop, ask a person rather than a model. It is easy to mock. But underneath the trend is what Leo is making in a more serious key, the claim that the effort was never the obstacle to understanding. The effort was the understanding.

The paradox he is naming is the one I find most worrying in my own practice and in the organisations I work with. AI rewards production, because production is what gets measured and valued. At the same time it erodes the slower thing, the learning, the burning curiosity, the willingness to sit with not-knowing, that made the production worth anything. We optimise for output and disinvest in the faculty that generates it. This is what I was getting at in Build to Think: the making is not separate from the thinking, it is the thinking. Hand the making to a machine and you do not just save time, you skip the part where you would have understood something.

He writes that educating people about AI "involves teaching them to decide when and for what purpose it ought not to be used." Not how to use it more. When to refuse it.

Which leaves the obvious problem of how anyone is supposed to decide. "When to refuse it" is a fine principle and a useless instruction, because the hard part is the deciding, in the moment, on a particular task, when the tool is right there and the deadline is not. There is some genuinely interesting work starting to form around this. One distinction separates offloading, where a tool supports your thinking, from outsourcing, where it replaces it, and argues the two have completely different consequences even when they look identical from the outside. Another line of research suggests the worst place to be is the scattered middle, using AI for a bit here and there, which adds the overhead of managing the tool without freeing enough capacity to think differently. The test I have landed on, and it is more a question I ask afterwards than a rule I apply before, is whether I understand something now that I did not understand an hour ago. If the answer is no, the speed bought me nothing. It is not a framework yet. But I suspect the thing worth building is closer to that question than to another list of prompts.

An error is not always a flaw

Leo draws a distinction between how a machine treats a mistake and how a person does. "For an algorithm, an error is a flaw to be corrected; for a person, however, an error can be a catalyst for profound change."

The systems we are building are optimisation engines. They are designed to reduce error, to converge on the correct output, to iron out the deviation. Much of the time that is exactly what we want. But a great deal of what makes us human, and a great deal of what makes innovation possible, lives in the deviation. The failed adhesive that became the Post-it Note. The wrong turn that opened the better road. The draft that revealed what you actually meant only because the first attempt was wrong.

A person's future, he writes, is not calculable; it depends on freedom and on the relationships a person cultivates. A system that classifies and optimises what already exists can, in his words, "however unintentionally, become an obstacle to change and growth." It is a more generous and more interesting critique than the usual one. The danger is not that AI makes mistakes. The danger is that it does not, and that we come to treat our own errors the way it treats its own: as flaws to be eliminated rather than openings to be followed.

But I want to be honest about my own reading here, because the error-as-catalyst line is the kind of sentence I am inclined to underline and move on. The truth is more uncomfortable. Most errors in most organisations are not catalysts for profound change. They are costly mistakes that hurt people, lose money, and absorb time that could have gone elsewhere. The Post-it Note is the story we tell because the other story, the one where the failed adhesive was just a failed adhesive, doesn't make it into the book. Used too quickly, "an error can be a catalyst" becomes a sentimental cover for not fixing things that should be fixed. The harder question Leo's distinction actually poses is which errors are which, and we are not very good at telling them apart in the moment.

Holding the tension

There is a red thread running through all four of these ideas. He sets the whole thing between two biblical images: Babel, the project built for its own grandeur that ends in dispersion, and Nehemiah rebuilding the walls of Jerusalem piece by piece, each person given their own stretch of wall, the relationships rebuilt before the stones. The choice he poses is not whether to use the technology. It is which of these two things we are building with it.

I am wary of tidy endings, and he resists one too. He does not tell us the tool is bad. He is explicit that technology is "a profoundly human reality" and that the creative intelligence behind it is a gift. What he asks is narrower and harder: that we keep asking whether what we are building makes human life "more human in every aspect," a question he borrows from John Paul II and aims squarely at AI.

My own hunch is that we are asking the wrong question when we ask how much to use these tools. Leo's phrase points somewhere else. The question is not how much we are producing but whether we are becoming more in the process, and that is a thing you can only ever see in retrospect, never at the moment you reach for the tool. What I am left with is a sharper version of the unease I started with. The tools are extraordinary. They also set a pace that is not ours, reward a kind of producing that hollows out the knowing, and treat error as a defect when sometimes errors can be a door to new discovery. Telling the doors from the defects, and not collapsing the distinction in either direction, seems to be the actual work.

So I will leave it where he leaves it, more or less. If an error can sometimes be a catalyst rather than a flaw, what in your own work might you currently be trying to optimise away that you would do better to explore? And, just as honestly, what are you calling exploration that you might do better to fix?