Would AI Choose the Cross?

Today we explores the fundamental limitations of Artificial Intelligence through the lens of Good Friday. The text Argues that while AI is the ultimate "optimization machine" designed for efficiency and self-preservation, it lacks the capacity for sacrifice—an act that is often "irrational" by algorithmic standards. By comparing "Machine Logic" (minimizing cost) with "Divine Logic" (choosing suffering out of love), the text concludes that the most essential human qualities—intention, soul, and self-giving love—are features that can never be programmed or outsourced to a model.

A Catholic technologist's most uncomfortable question

It's Good Friday. The office is quiet. The coffee is strong. And I'm sitting here — a Catholic tech entrepreneur in the Netherlands, somewhere between a Hail Mary and a pull request — asking myself the most uncomfortable question I've had all week: If we built an AI smart enough to understand sacrifice, would it ever choose it?

Spoiler alert: No. And that's kind of the whole point.

Look, I've spent the last year building AI solutions for clients who barely know what a token is. I've watched language models write emails, generate reports, optimize logistics routes, and occasionally hallucinate things that would make a priest blush. But today — today — I want to talk about what AI cannot do. Not because of a technical limitation. Because of something far more fundamental.

The Most Rational Being in the Room

Here's the thing about AI: it is, at its core, a glorified optimization machine. Ask it to solve a problem and it will find the most efficient path. Ask it to minimize harm and it will calculate trade-offs with the cold precision of an accountant who's never had his heart broken. It is, in many ways, the most rational being in any room.

And that, my friends, is exactly its problem.

Because rationality has its limits. And Good Friday is the day we remember that the greatest act in human history was, by any algorithmic standard, completely irrational.

"No AI would walk willingly into Gethsemane. It would find a workaround. It would optimize the outcome. It would file a bug report titled: 'Crucifixion: High severity, affects primary user.'"

Let's Run the Prompt

Imagine you give a sufficiently advanced AI the following scenario. You are all-knowing, all-powerful, and you love humanity. Humanity is broken. You can fix it — but only by suffering completely: betrayal, humiliation, pain, death — while having the full ability to stop it at any moment. What do you do?

A well-aligned AI would respond something like: "I have identified 47 alternative solutions with lower personal cost and equivalent outcomes. Shall I rank them by efficiency?"

Jesus said: "Not my will, but yours."

That's not a logic error. That's love. And no loss function in the world was built to compute it.

AI Logic vs. The Cross — Side by Side

Machine Logic	Divine Logic
Maximize self-preservation	Full knowledge. Full power. Full choice.
Minimize own suffering	Chose suffering anyway
Find the optimal, least costly path	Not the efficient path — the right one
Escalate if constraints are violated	Forgives while being killed
No concept of "dying for another"	Turns death into the greatest gift
Love = undefined variable	Love = the entire algorithm

The Bug That Isn't a Bug

In AI alignment circles, one of the most debated topics is how to make systems that are truly beneficial — not just efficient, not just accurate, but genuinely good. It's harder than it sounds. Because goodness, real goodness, sometimes looks like loss. It looks like giving up the easier road. It looks like a Friday afternoon on a hill outside Jerusalem.

We can train a model on every theological text ever written. Feed it the Summa Theologica, the Desert Fathers, the entire Catholic Catechism. It will summarize them beautifully. It will pass any theology exam you throw at it.

But it will not understand sacrifice. Because sacrifice isn't information — it's intention. It's the willingness to absorb pain so someone else doesn't have to.

That requires a soul. And as far as I can tell — and I've checked the documentation pretty carefully — we have not shipped that feature yet.

"An AI can be trained to say 'I love you.' It cannot be trained to mean it when it costs everything."

What This Means for Those of Us Building With AI

I'm not writing this to bash the technology I work with every day. AI is extraordinary. I've seen it change how small businesses operate, how teams communicate, how ideas scale. It's a remarkable tool.

But a tool is not a savior.

And on Good Friday, more than any other day, that distinction matters.

What we're building in the AI space is powerful, but it is morally neutral until a human being — one with values, history, and yes, faith — gives it direction. The ethics don't live in the model. They live in us. We are the ones who decide what it's used for. We are the ones responsible for its impact. That responsibility is not something you can outsource to a language model.

Jesus didn't outsource the cross. He carried it.

A Closing Thought, Before the 3 O'Clock Bell

There's a reason Good Friday is called good. Not because it was pleasant — it was, by all accounts, horrific. But because what came from it was the greatest reversal in history: death becoming life, loss becoming redemption, the most human moment becoming the most divine.

AI will never have a Good Friday. It will never face a moment where it must choose between its own continuity and the good of another. It doesn't have skin in the game.

It doesn't have skin at all.

But we do. And maybe that's the reminder this day is really giving us: that the things that make us most human — our capacity for sacrifice, for love that costs something, for faith that doesn't always compute — are precisely the things that no model, no matter how large, will ever replace.

"God didn't send an algorithm. He sent His Son."

Feliz Viernes Santo. From The Netherlands, with faith and a GPU.

✝ ✝ ✝