Start with something familiar
Your phone sorts photographs. An email service flags spam. A chat assistant turns a messy note into a tidy plan. These are different jobs, but all can involve artificial intelligence: computer systems built to perform tasks such as recognizing patterns, making predictions, generating material, or choosing actions.
AI is an umbrella term, not one product. A system that predicts tomorrow’s demand is not necessarily a chatbot. A tool that draws an image may work differently from one that sorts customer messages. Ask what a particular system does and what evidence would show that it does the job well.
Where an LLM fits
A large language model, or LLM, learns patterns from large amounts of training material. At use time it generates text in pieces called tokens. That process can support useful writing, coding, summarizing, and problem solving. It can also produce a plausible sentence that is factually false.
Some modern models accept images, audio, or other inputs too. We call these multimodal models. A chat app may also combine several models and external tools, so the whole app can do more than a single model on its own.
The skill worth learning first
Treat an answer as a proposal you can inspect. If you supplied five facts, check that the response preserved all five. If the assistant introduces a new claim, ask where it came from, open the source, and check that the source supports it. Smooth wording cannot tell you whether a claim was checked.
Your five-minute experiment
Write three fictional errands: return a library book, buy potting soil, and pick up a repaired bicycle. Ask an assistant to make a checklist using only your list. Then ask what information it would need to put the errands in a sensible order. You are learning to give context and recognize missing information without needing technical vocabulary.