Prompt
In short
A prompt is the input text or instructions you give an AI model, such as an LLM, to tell it what task to perform and what kind of answer you want.
What is a prompt?
A prompt is everything you send to a language model before it starts generating a response. It can be a simple question, a detailed set of instructions, examples of the desired output, or pasted documents for the model to work with. The answer depends heavily on what the prompt contains and how clearly it is written.
Many AI APIs split a prompt into messages with different roles. A system prompt sets the overall behavior and rules, such as tone or output format, while user messages contain the actual requests. Including a few worked examples in the prompt is called few-shot prompting, while asking with no examples is called zero-shot prompting.
Writing a prompt is a lot like briefing a new colleague: the more context, constraints, and examples you give, the less they have to guess. Good prompts state the goal, provide relevant background, specify the output format (for example, JSON with certain fields), and say what to do when information is missing.
Prompt engineering is the practice of designing and testing prompts to get reliable results. It is different from training or fine-tuning: a prompt only changes the model's behavior for that request and does not change the model itself.
Key takeaways
- A prompt is the input that tells an AI model what to do.
- System prompts set the rules; user prompts carry the specific request.
- Clear instructions, context, and examples lead to better answers.
- A prompt affects a single request and does not retrain the model.
Example
// Rules go in the system message, the actual task in the user message
const messages = [
{
role: "system",
content: "You are a code reviewer. Reply in JSON with the fields 'issues' and 'summary'.",
},
{
role: "user",
content: "Review this function: function add(a, b) { return a - b; }",
},
];
// Send the whole prompt to the model (callModel is a placeholder)
const reply = await callModel(messages);Readers ask
What is prompt engineering?
Prompt engineering is the practice of writing, testing, and refining prompts so an AI model produces accurate, consistent, and useful output. It includes techniques such as giving clear instructions, adding examples, and requiring a specific output format.
What is a system prompt?
A system prompt is a special instruction, usually hidden from the end user, that sets the model's role, rules, and style for a whole conversation. User messages then add specific requests on top of it.
What is prompt injection?
Prompt injection is an attack where untrusted text, such as a web page or user input, contains instructions that trick the model into ignoring its original rules. It is similar in spirit to SQL injection, so applications should treat any outside text sent to a model as untrusted.
See also
- LLMAI & Machine Learning, p. 25An LLM is a machine learning model trained on huge amounts of text that generates language by repeatedly predicting the next most likely piece of text.
- RAGAI & Machine Learning, p. 38RAG is a technique that makes an LLM answer using relevant documents retrieved at question time, so its responses are grounded in current, specific data.
- HallucinationAI & Machine Learning, p. 23A hallucination is when an AI model, such as an LLM, confidently produces information that sounds plausible but is false, invented, or unsupported by sources.
- Machine LearningAI & Machine Learning, p. 27Machine learning is a branch of artificial intelligence in which computers learn patterns from data to make predictions instead of following hand-written rules.
- SQL InjectionSecurity, p. 40SQL injection is an attack where user input is treated as part of a database query, letting an attacker read, change, or delete data they should not reach.
- Context WindowAI & Machine Learning, p. 12A context window is the maximum amount of text, measured in tokens, that an LLM can consider at once, including the prompt, conversation history, and its reply.
- System PromptAI & Machine Learning, p. 44A system prompt is the instructions an app gives a language model before the conversation starts, setting its role, rules, tone and what it should know.
- Prompt InjectionSecurity, p. 29Prompt injection is an attack on LLM apps where attacker-written text is treated as instructions, so the model ignores its rules, leaks data or misuses tools.
Spotted a mistake or something missing on this page?Suggest an edit