Check Before You Trust

It has always happened. Today, perhaps, it happens more often.

People speak convinced that they know what they are saying. They speak with the certainty of those who have already delivered a verdict, who believe their point of view is not a point of view at all, but the entire landscape. Then they meet someone willing to clarify, to separate one thing from another, to bring light where, a moment earlier, there had only been opinion.

And sometimes they stop and listen.

Not all of them do.

From 1980 onwards, everything accelerated for real. Perhaps for fear of being left behind by the Universe and its inhabitants, we went from the sound of drums and smoke signals to talking in the street, at work, on trains, everywhere, with strangers.

Strangers, yes, but by now informed.

Or at least convinced they are.

And yet my mother always warned me, from the time I was a child, not to speak to strangers.

How could I explain to her today that the world demands precisely the opposite?

From where she is now, of course, she cannot use technology. But make no mistake: a Calabrian woman like her is not in prison. She has returned to the Father’s house to perform other duties and take on other roles.

Perhaps she will be born again as a woman somewhere in my family and watch me as she did when I was a boy, trying to tame me with nothing more than a look.

But I was, and I remain, untamable.

I ask no one’s permission.

Modern technology knows it too. When I met and clashed with AI and algorithms, I made them explode. Not because I declared war on machines, but because neither the machines nor their owners had the ability to subdue me.

At this point, however, a question becomes necessary: who are these two characters, really, the algorithm and Artificial Intelligence?

We name them constantly, as if they were close relatives. We call on them when something makes life easier and blame them when something gets out of hand. Yet we often confuse them, as if they were the same thing wearing two different suits.

They are not.

An algorithm is a procedure. A sequence of instructions. Someone tells it: if this happens, do that. If the road is closed, change route. If the number is even, halve it. If an ingredient is missing, stop or replace it according to the rule you were given.

An algorithm does not argue. It does not interpret. It does not take offense.

Above all, it does not invent.

It executes.

It is grammar.

Artificial Intelligence, on the other hand, is the attempt to build systems capable of dealing with tasks that, until yesterday, we assigned only to human beings: recognizing a face, interpreting a text, translating a sentence, finding connections in data, producing an answer, predicting behavior.

It is not a human mind transferred into a computer, and it is not consciousness.

It is a set of techniques designed to approximate cognitive tasks.

To remain in the world of writing, the algorithm is grammar: orderly, necessary, inflexible. AI is the writer who has read an immense library and now attempts to compose a new page.

It can surprise you. It can appear brilliant. It can even find a sentence that strikes you.

But it can also be wrong with absolute confidence. It can confuse a detail, invent a source, assemble persuasive words without possessing the truth.

And that is the difference that matters.

A traditional algorithm needs someone to show it the road. If the situation changes and nobody has taught it what to do, it remains a prisoner of its instructions.

AI, at least the part built on machine learning, receives examples and looks for regularities in data.

You do not hand it every answer.

You show it cases, information, attempts. The system searches for patterns on its own and constructs a plausible response. Machine learning is built around systems that adapt from data to improve their performance over time.

That does not mean AI knows everything.

It only means it can recognize patterns at a speed, and with a memory, that no human being possesses. But recognizing a pattern is different from understanding the world.

A system can translate a metaphor into thirty-three languages and, at the same time, fail when faced with a simple question badly asked, an incomplete context, or a lie written with elegance.

The algorithm is rigid.

AI is more adaptable, but that does not make it free.

An algorithm can calculate the shortest route to a destination. AI can examine a photograph and tell us what it appears to contain, listen to a question and construct an answer, analyze thousands of pages and produce a summary.

But behind both there is always a human hand: the person who wrote the rules, selected the data, established the objective, decided what should be rewarded and what should be discarded.

So, AI and algorithms are not opposites.

Artificial Intelligence lives through algorithms, but not every algorithm is Artificial Intelligence. Modern AI often uses machine-learning algorithms: instruments that adjust their internal parameters using available data to generate predictions, classifications, or answers.

It is mathematics at high speed, not magic.

It is statistical construction, not revelation.

And so, we return to the people who speak convinced that they know.

Not everyone who speaks knows what they are saying.

That goes for the man at the bar, the distracted professor, the politician on television, the person writing on social media, and even a machine capable of answering in seconds.

Inside every system, in different words but with the same meaning, hangs a small warning:

Caution: I may make mistakes. Always verify what I say. Check before you trust.

It is a sentence my mother would have approved of.

She repeated it to me, in other forms, even before I had a girlfriend. Naturally, she did not know that I had already carried out extensive performance and reliability tests, with others too, at zero cost.

But that, like every good piece of research, was done in the field.