Artificial intelligence is a broad category of systems designed to perform tasks associated with intelligence. Machine learning learns patterns from examples rather than requiring every rule to be handwritten. Deep learning uses neural networks with multiple layers. Generative AI produces outputs such as text, images, or audio; a large language model is one kind of generative model focused on language-like sequences.
These terms overlap but are not interchangeable. A spam classifier predicts a category. A text generator produces a sequence. A retrieval system finds existing evidence. A product can combine all three, and each component needs a different evaluation.
A model has parameters learned during training. Inference uses the trained model on new inputs. Application context supplies information at runtime without necessarily changing those parameters. Keeping those distinctions clear helps you choose between adding evidence, improving instructions, and adapting a model.
Exercise: classify a search engine, an invoice extractor, a ticket classifier, and a drafting assistant by their main input and output behavior.
Check: you can explain why retrieving a document is different from training a model on it.