AI has the recipe, but you have the fricandó touch.

This weekend I put on an apron to cook a fricandó. It’s a dish I know well, but even so, I always look up a recipe before I start. Why?

It’s not because I don’t know how. I do it to confirm, to compare, and above all, to learn from other people’s nuances. But there’s one crucial detail: I always look for a recipe from someone who has actually cooked it, I look for their real experience.

If you ask an AI for a fricandó recipe, it will give you an impeccable summary of everything it has read on the internet. But that summary is sterile. AI hasn’t smelled the sofrito, and it doesn’t know the exact moment when the onion hits that perfect caramelized point that makes the difference between a decent dish and an outstanding one.

AI doesn’t have experience, it has patterns. And where the pattern doesn’t reach, AI hallucinates and invents, and in that “perfect” answer it might miss something as essential as the onion.

The same thing happens with code.

If we ask an AI to generate something complex, the structure might look “perfect”, but something essential can be missing – something only practice teaches you.

AI doesn’t know what it doesn’t know. It lacks that global view to understand that a solution can be syntactically perfect but semantically a disaster. It can’t foresee a context bug that a developer would catch by pure “gut feeling”.

Blind trust is the danger. Programming using only AI, without applying our judgment, is like trying to cook fricandó without onion: technically it may be a stew, but it’s not what it should be.

AI is an incredible tool, but the real value still lies in the person behind it. The value is in that professional who reads the “recipe” (or the code), but brings their experience, their ability to correct, and their personal touch.

Let’s not delegate our judgment. Let’s use technology to add value, and always make sure that, in the end, the fricandó has onion. 🧅🥘

#AI #Development #Programming #vibecode #CleanCode