How does machine learning learn?

Machine learning algorithms identify patterns in training data to make predictions on new data. 'Learning' means adjusting internal parameters to minimize prediction errors. It requires large datasets, computational power, and careful design to avoid learning spurious correlations.