How does AI learn?

Welcome! I am excited to explore this with you.

To help me tailor this session for you, what background or experience do you have with computer science or AI, and is there a specific goal or practical application you'd like to understand best?

I'm a complete beginner looking for an intuitive overview.

Welcome! I am glad to have you here. Since you are looking for an intuitive overview, we will build things step by step without any complicated jargon.

Your path

The Three Main Types of AI Learning

How Neural Networks Work (Deep Learning)

Real-World Applications & Ethical Considerations

To start off, let's look at how computers used to solve problems versus how they do it now with AI.

In traditional computer programming, a human programmer writes explicit line by line. For example, if you wanted a program to filter spam emails, you might write a rule like: "If the subject contains 'WINNER' or 'FREE', mark it as spam." The computer simply follows those hardcoded instructions without thinking.

flips this process completely upside down. Instead of handing the computer rules, you feed it thousands of examples of regular emails and spam emails—this is called .

The system analyzes all those examples, spots subtle patterns on its own (like combinations of words, senders, or formatting), and builds its own internal set of guidelines to identify spam in the future.

Continue to training data as the foundation for AI

Exactly! Think of training data as the textbook or flashcards for the AI.

If you give the system ten thousand examples of cat photos and dog photos, it learns to look for patterns—like ear shape or snout length—without anyone explicitly writing code for those features. The more high-quality examples it sees, the better its predictions get.

Now that we've covered the basics of learning from data, let's explore the three main ways AI actually learns: Supervised, Unsupervised, and Reinforcement learning.