Tag: <span>ETHICS</span>

AI didn’t force its way in. It made itself indispensable through convenience. And we let it in, without really measuring what we were giving up in return.

In May 2026, Pope Leo XIV published Magnifica Humanitas. An encyclical about AI. Yes, you read that right. And this text says, with a clarity that most tech reports lack, what we still refuse to face: human dignity eroded by algorithms, thought that offloads its own effort, power concentrating in the hands of those who code while everyone else just absorbs the consequences.

I read this text as a practitioner, not a believer. And what I found in it mirrors exactly what I observe every day: we are no longer just automating tasks. We are delegating the way we think.

Between Babel and Jerusalem, the encyclical asks a simple, almost uncomfortable question: which construction site are we part of, right now, with the tools we use every single day? What are we actually building? And for whom?

A machine was fed with the equivalent of 10 million books. Then it was taught humanity.

he result? GPT-3 began solving advanced mathematics. Without anyone teaching it. It translated between programming languages. Detected invisible emotional signals. Wrote sonnets under seemingly impossible constraints.

Then GPT-4.5 arrived. Worse than its predecessor on certain tests. With no obvious explanation.

As if these models had their own cognitive weather.

Between 2018 and 2022, behind the closed doors of research laboratories, something was created that we still do not fully understand. Something that develops abilities no one explicitly taught it. And sometimes loses them, without anyone knowing why.

The cost of the operation: 200 million dollars. 25,000 GPUs. 20,000 hours of human labor just to teach it basic politeness.

If their creators discover abilities only after the fact, what other surprises might still be waiting for us?

An orchestra without a conductor is organized chaos. Each musician plays in tune, yet together, things begin to drift. AI works the same way. It produces striking text while being able to slip at any sentence, all with the same confident tone. For a long time, we believed there were only two options: speak to it better (prompt engineering) or reprogram it (fine-tuning). But there is a third path, more subtle. A way of whispering directly into its internal states, while it is thinking.

This is called steering.
And it changes everything, not only our relationship with AI.
Our relationship with ourselves as well.