How we use AI

We use it, quite a lot. We would rather tell you where and why than let you guess.

Where a page starts

With someone who remembers not understanding. Every topic here grew out of a question we actually asked ourselves: why don't the planets fall into the Sun, what happens inside a battery, why is the night sky dark if there are infinite stars. Deciding which question deserves a page, where to enter it from, which image makes the idea click and which one quietly confuses: nobody delegates that part. That is the craft.

Translation, and this is where AI earns its place

We are a small team, and the site speaks three languages, each at three levels of depth. Nine versions of the same idea, every one in its own register: a curious ten-year-old does not read like a graduate student, and a badly translated text gives itself away instantly. It sounds foreign, and it pushes people away.

Without AI we would have had to choose: one language done well, or three done badly. With it we can hold the same standard of language across all three, the same care in how a sentence lands, the same precision in technical terms. A kid in Lyon and a kid in Bologna each read a page written in their own language, not a translation that creaks. That matters enormously to us, and it is the main reason AI is part of this project at all.

Explanations and simulations

Draft text and the code behind the interactive simulations start with AI. Then we read them, correct them and pull them back into line, because there is one thing we won't bend on: a wrong scientific explanation does more damage than no explanation. A simulation that lies is worse than a still drawing. If a number doesn't add up, or a behaviour isn't physically honest, the page does not ship.

What always stays with us

Choosing the topics, the angle, checking the facts, deciding to publish, and taking responsibility for what you read. AI is a tool we work with, not a signature at the bottom of the page.

Why we're telling you

Because Article 50(4) of Regulation (EU) 2024/1689, the AI Act, requires it: anyone publishing AI-generated text that informs the public on matters of public interest, and science is one of them, has to say so clearly. But also because it seems like the least we could do. A site that explains how things work cannot be cagey about how it works itself.

Found a mistake?

Write to [email protected] and we'll fix it. Reader corrections are our favourite kind of quality control.

The LabInATab team