What large language models are

A large language model (LLM) is a program trained on a vast amount of text to predict the next word. It sounds modest, but from this simple principle grows the ability to hold a conversation, write code, translate, summarize and reason.

Imagine the model has read a giant library — books, articles, forums, documentation — and learned to notice which words usually follow which. When you ask a question, it doesn't "recall" a specific page; it assembles the most plausible continuation of your text on the fly.

Think of the model as an extremely well-read companion who answers from memory — fast, confident, but sometimes wrong.

The beginner's biggest misconception

Many assume AI "searches the internet" or "stores a database of facts". It doesn't. By default the model isn't connected to the web at answer time and doesn't consult a database — it relies only on patterns learned during training. Two consequences to accept right away:

• Knowledge has a "cut-off date". The model knows the world roughly up to when it was trained. About fresh events it may not know, or may only guess.
• Confidence ≠ correctness. The model sounds equally sure whether it's right or wrong. Fact-checking is on you (a whole lesson on this later).

What makes up an answer

Every answer is assembled from three sources:

• Your prompt — what you wrote just now. The most controllable factor.
• Context — previous messages in this dialogue. The model "sees" and uses them.
• The model's training — the "experience" baked into its parameters. You can't change it, but you can steer it with your prompt.

Let's see it in practice

Same topic, two different prompts — a completely different result:

The difference isn't the model's "knowledge" but how much useful context you gave it. That's the skill we'll build all course long.

Exercise

Open any assistant and ask the same thing two ways: first as a single word ("Vacation"), then in detail ("Help me choose where to go on vacation this autumn for a week, mid-range budget, I love the sea and calm"). Compare the answers and feel how detail changes usefulness.

Remember: the model works with what you wrote, not what you meant. Want a better answer — give more context.