Supervised learning trains models on labeled examples, inputs paired with correct outputs, so they learn to predict labels for new inputs: spam or not, price to
Training models on labeled examples, inputs paired with correct outputs, so they learn to predict labels for new inputs: spam or not, price tomorrow, disease present. Label quality bounds everything it produces.
Fine-tuning is supervised learning applied to a pre-trained model: instruction tuning, for example, is supervised learning on prompt-response pairs. Supervised learning also remains the workhorse of business ML like scoring, classification, and forecasting.