How does AI learn?
When we say that AI “learns,” what is really happening?
Let’s go back to our analogy with the human brain.
When the human brain learns, what happens, in simplified terms, is that the neurons that were connected and talking to each other in one way adjust and begin to talk in a slightly different way, or very differently, depending on the intensity of training or experience.
With the AI neural network, the process is very similar.
The little “pieces” that are talking to each other begin to talk in a different way after training.
So we can say that we are really imitating how the human brain learns in AI learning.
And the amazing thing is that it’s working, even without fully understanding the human brain or the AI “brain.”
What matters, at the end of the day, are the results.
And these results can be achieved through different types of training, just like human learning.
Light training for AI can be like giving an employee a small piece of feedback, gradually adjusting their behavior.
Intensive training, on the other hand, can be compared to sending an employee to do a 3-year MBA, promoting a profound change in how they work.
Understanding these different levels of training will be fundamental for managing AI agents effectively, as we’ll see later in this book.