What a large language model is

Think of the autocomplete on your phone: you type "See you" and it suggests "tomorrow". A large language model (LLM) is the same idea scaled up enormously. Given some text, it predicts what text should come next, adds that, and repeats until the answer is complete.

To learn this, the model was trained: a computer program read a huge amount of text (books, websites, code, conversations) and slowly adjusted billions of internal numbers until its predictions got good. "Large" refers to that size. After this first stage, the model is trained further on examples of helpful conversations so that it answers questions and follows instructions instead of just continuing your sentence.

Chat apps such as ChatGPT (from OpenAI), Claude (from Anthropic) and Gemini (from Google) are built on top of LLMs. There are also open-weight models (models whose files anyone can download and run, such as Llama, Mistral, Qwen or Gemma). The same models power AI steps inside tools like Zapier, Gumloop and n8n.

What this means in practice:

• The model is very good at language tasks: drafting, summarising, translating, classifying, extracting facts from text, and writing code.
• It does not look things up unless it has been connected to a search or a database. By default it writes from patterns it learned during training.
• It has a knowledge cutoff: the date its training text ends. It does not know about later events unless you tell it.
• Its answer is generated fresh each time, so the wording can change from one try to the next.

The text you give the model is called the prompt. Everything else in this course is about writing good prompts and connecting models to the rest of your work.