What is AI, anyway? A cake recipe explains

Imagine you want to teach someone to bake a cake.

The traditional way is to hand over the recipe: 3 eggs, 2 cups of flour, oven at 350 degrees. The person follows it step by step. That is how most computer programs work: someone writes the rules, the computer obeys.

Artificial intelligence does it differently. Instead of the recipe, you show it thousands of finished cakes and say: "this one turned out well, this one badly, this one sank". Over time, it spots the patterns by itself: "when there are too few eggs, it sinks".

So does AI think?

No. It recognizes patterns very well. It is like that friend who guesses the ending of any movie because they have watched so many. They are not psychic, they have just seen a lot.

Why this matters to you

Because it explains both sides of AI:

• The good side: it becomes a champion at repetitive tasks, such as reading scans, translating texts or suggesting songs.
• The side that needs care: if the examples it saw were bad or incomplete, it learns wrong. And it learns wrong with great confidence.

Keep this

AI does not follow a recipe. It learns by looking at examples. That is why the quality of the examples is everything.

In the next edition of the series, we tell you where those examples come from. Spoiler: many came from you.