What is Prompt Engineering?
Prompt engineering is the craft of designing inputs that reliably steer a large language model (LLM) toward useful, accurate, and safe outputs. It is fast becoming a core literacy for any knowledge worker — and especially for public servants who must balance speed with accountability.
Why it matters for civil servants
AI tools like ChatGPT, Gemini, and Claude are now used in policy drafting, citizen communications, translation, and summarising long reports. The quality of what you get out depends almost entirely on the quality of the prompt you give in. A poor prompt wastes time and can produce misleading content; a strong prompt turns a generic assistant into a focused colleague who works at the speed of typing.
Mental model: the brilliant but context-free intern
Think of an LLM as an extremely well-read but context-free intern. It has read most of the public internet, but it knows nothing about you, your ministry, today's date, or what 'good' looks like in your office — unless you tell it. Every prompt is essentially an onboarding briefing for that intern, compressed into a few sentences.
How LLMs actually generate text
LLMs predict the next token (a word fragment) based on everything that came before. They do not 'look things up' — they generate the statistically most likely continuation of your prompt. This explains both their fluency and their failure modes: they will confidently produce something that sounds right even when it is wrong.
Three things prompt engineering is NOT
- It is not magic incantations — there are no secret words that unlock hidden capabilities.
- It is not programming — you don't need to learn a syntax; you need to learn how to communicate clearly and specifically.
- It is not a one-time skill — models change, and the prompts that worked last year may need updating for new versions.
📖 Key terms
- LLM —
- Large Language Model — an AI system trained on huge text corpora to predict and generate language.
- Prompt —
- The text input you give an AI model to elicit a response.
- Token —
- A small chunk of text (roughly 3–4 characters) that models read and produce one at a time.
- Context window —
- The maximum amount of text (in tokens) the model can consider at once, including your prompt and its reply.
🎯 Check your understanding
Answer all questions, then submit to see your score.