AI explained

LLM meaning: what a large language model is, in plain English

In artificial intelligence, LLM means large language model: the technology behind ChatGPT, Claude, Gemini and most AI assistants. In South Africa the letters also mean a Master of Laws degree, which is why the question comes up so often. Here is what an LLM is, how it works, what it does well and badly, and what it means for a business.

Chad AlexanderCo-founder and AI engineerUpdated 28 September 2026

The position

A large language model is software trained on enormous amounts of text to predict the next word, and that simple skill turns out to be enough to write, summarise, translate, answer questions and write code. It does not look facts up unless it is connected to search or to your documents, which is why it can sound confident and still be wrong.

For a business, the useful question is not "which LLM" but "what job": answering customers, reading documents, drafting quotes. The model is a component; the value is in connecting it to your data and your process. That is the work we do; see AI engineering.

What LLM stands for, in AI and in law

In AI, LLM stands for large language model. "Large" refers to the size of the model, billions of internal settings learned during training, and to the size of the text it learned from. "Language model" means its core skill is working with language.

In South African universities, LLM also stands for Legum Magister, the Master of Laws postgraduate degree offered by universities such as UNISA and Wits. If you arrived here looking for the law degree, that is the other LLM; the rest of this page is about AI.

How a large language model works

The short version, without the maths.

01
Training on text
The model reads a huge collection of text and learns patterns: which words and ideas tend to follow which.
02
Predicting the next word
Given some text, it predicts what comes next, one small piece (a token) at a time. Long answers are built from many of these predictions.
03
Tuning for helpfulness
After training, it is tuned with human feedback so it follows instructions and answers helpfully and safely.
04
Connecting to tools
Modern assistants connect the model to search, documents and apps, so answers can use current facts instead of memory alone.

What LLMs are good and bad at

Writing

Good at
Drafting emails, quotes, summaries and reports quickly.

Watch out for
Confident errors if nobody checks the facts.

Reading

Good at
Pulling key terms, dates and figures from long documents.

Watch out for
Missing context that is not in the text it was given.

Answering

Good at
Answering common customer questions instantly, any hour.

Watch out for
Inventing answers when it has no source: always ground it in your data.

Coding

Good at
Writing and fixing code, fast.

Watch out for
Subtle bugs that need a developer to review.

Maths and facts

Good at
Explaining concepts and reasoning step by step.

Watch out for
Exact figures, unless it uses a calculator or a trusted source.

LLM terms you will hear

Prompt

The instruction or question you give the model.

Basics

Token

A small chunk of text, roughly three quarters of a word, that models read and write and that APIs charge for.

Basics

Context window

How much text the model can consider at once, from a page to whole books.

Capacity

RAG

Retrieval-augmented generation: the model looks things up in your documents before answering, which cuts made-up answers.

Business

Agent (agentic AI)

An AI that takes actions, such as sending messages, updating systems or running steps on its own, not just answering.

Business

Hallucination

A confident answer that is wrong or invented.

Risk

What it means for your business

You do not need to build an LLM. You need to aim one at a real job and connect it to your data safely, with POPIA in mind and a person checking anything that matters. See ChatGPT versus a custom AI agent and whether AI is worth it for a South African business.

LLM meaning: common questions

What does LLM mean in AI?

Large language model: an AI trained on large amounts of text to understand and generate language. ChatGPT, Claude and Gemini are built on LLMs.

What does LLM mean in law?

Master of Laws (Legum Magister), a postgraduate law degree offered by South African universities including UNISA and Wits.

Is ChatGPT an LLM?

ChatGPT is an assistant built on OpenAI's large language models, with extra tools such as search, file handling and image generation added around the model.

What is AI in Afrikaans?

Kunsmatige intelligensie, often shortened to KI. A large language model is a groot taalmodel.

Can a business use an LLM safely?

Yes, if it is connected to the right data, kept inside controlled accounts, and checked by a person for anything involving money, law or health. POPIA applies to any personal information it handles.

Have a job an LLM could do?

Tell us the task, and we will say whether AI is the right tool and quote one fixed price in rand to build it properly.

Start the build→