Toward desirable futures
Thomas di Luccio
Mistral AI | The Algorithms of Escape

Mistral AI | The Algorithms of Escape

The Algorithms of Escape

Thomas di Luccio • February 1, 2026

I'm on a lifelong journey toward emancipation.

Emancipation from the fear that I'm not enough. From the labels and roles designed to keep me small.

The more I travel through life, the more I see the opportunities. The ones I missed because I hesitated. The ones I could still take, if I dared. I know most of this work is mine to do.

I sometimes need a little help to feel enough. This is pushing me to learn more, try new things, master new skills, and create new tools.

I don't know how healthy that is, really, but I live with this voice urging me to have a backup plan. Always.

What's next? What's my move if things go south? What job could I do? What company could I start? Or join? How to burn toxic ones to the ground?

I don't have a clue, most of the time. But I live in a state of hyperawareness, the way people do when they've been ambushed by life once too often.

In that context, I see the rise of artificial intelligence (AI) as a beacon of hope. Use it right, and suddenly, the walls feel a little lower. The locks a little easier to pick.

I could not think of anything more suited to my anxiety and insecurities. I strongly believe my best asset is that, given enough time and trust, I can learn my way out of almost anything.

I have this will, this force. And now, I have this tool to fast-track.

But.

Of course, there is a but. There is always one.

Saying that the immensity of the resources needed to train all those models is a challenge. That's an understatement. It's a one-way ticket to a world on fire.

We are sold a world free of the suffering of labour. What we're getting is a world with fewer jobs, more slop, and the same old greed. Just automated.

AI is not stealing jobs. The people who treat labour as disposable are. They don’t know they are shooting themselves in the foot. They don't care who else they take down with them

How long before I'm on the line? Do I wait for the axe to fall? Or do I grab the wheel?

How much can I learn before the floor drops out? Could I build tools that actually serve me. Not just as a safety net, but as a trampoline? Could it also serve as plan B should the room left for human professionals shrink to nothing?

In that sense, you can feel the hiatus. On one hand, AI is a liberating force for individuals. On the other hand, it is this dooming one testing as rarely before the planetary limits.

This is yet another example of the opposition between the granular individual gesture that can be made by the willing and the collective framework that should be put in place to ensure we all stay within the average rise in temperature, allowing the planet's habitability.

There is social pressure to adopt the former, and we have yet to see anything move on the latter.

Here's the thing about building algorithms of escape: They require raw material.

My "skills" are trained on datasets scraped from underpaid workers.

My "efficiency" runs on servers cooled by melting ice caps. Even my backup plans have a carbon footprint.

I tell myself I'm just one person. What's one more script in the noise? But that's the lie we've been sold: that individual action is either powerless or sufficient. The truth is in the tension. The systems that trained these models also trained us. To optimise, to extract, to call it "innovation" when it's just kicking the can down a crumbling road.

I don't know if my tools will save me. But I know this:

The same systems that trained these models on our collective exhaustion also trained us to adapt when the floor gives way, to resist when the defaults are toxic, to find the cracks in the wall and pry them open with our bare hands.

Maybe that's the real skill. Not prompt engineering, not growth hacking, but learning to live, and build, in the rupture.

So I'll keep building.

But not for the myth of self-sufficiency, or the delusion of a solo escape. I'll build for the people who've been here before me, who left handholds in the dark. I'll build for the ones coming after, who'll need those cracks to widen into doors.

The algorithms of escape won't free us. But the right ones might buy us time. To organise, to grieve, to imagine a future less lonely than a backup plan

So what's a builder to do?

Keep building, but differently.

With a moral compass, yes. With a spine. But also with this:

The stubborn belief that the cracks we make today become the doors of tomorrow.