Many AI agents learn to make decisions by being trained on large amounts of data, a process known as machine learning.

Imagine you want to teach a child to recognize a cat. You wouldn't just give them a list of rules like 'it has fur and four legs'. Instead, you'd show them many pictures of cats, saying 'that's a cat,' and also pictures of dogs or birds, saying 'that's not a cat.' Over time, they'd learn to spot a cat on their own. Machine learning works similarly for computers. Instead of giving an AI agent every single rule it needs, you feed it tons of examples – called 'data.' The computer then looks for patterns in this data and uses those patterns to build its own set of rules or models. This way, when it sees something new, it can use what it learned to make a decision, like identifying an email as spam or suggesting a product you might like.