1.2 — Applications of LLMs
Where LLMs are used today — translation, summarisation, chatbots, code — and why their reach keeps growing.
Because LLMs are so good at parsing and producing unstructured text, they slot into a huge spread of real-world tasks.
The everyday toolbox
- Machine translation — moving sentences between languages.
- Text generation — writing emails, articles, stories, or even poetry on demand.
- Sentiment analysis — telling whether a review is positive or negative.
- Summarisation — boiling a long document down to its key points.
- Question answering — pulling a specific answer out of a pile of text.
- Content creation — drafting blog posts, marketing copy, or computer code.
- Chatbots & virtual assistants — ChatGPT, Google's Gemini, and the new wave of AI helpers baked into search engines like Google Search and Microsoft Bing.
Specialist knowledge retrieval
LLMs shine in narrow, technical domains too. In medicine or law, a model can sift through huge volumes of documents, summarise lengthy passages, and answer technical questions on demand.
Why this matters for you
Beyond simply using an LLM, these notes walk you through coding one from scratch — a model that can generate text — and then teaching it to follow instructions, summarise, translate, and answer questions.
In other words, you will build your own small ChatGPT-style assistant, step by step.
The rest of this chapter lays the conceptual groundwork so that, when we start writing PyTorch code in Chapter 2, you already know why each piece exists and what role it plays.