AI glossary · Beginner · Language models
Prompt
A prompt is the input that tells an AI what you want it to do. Questions, instructions, examples and reference material can all be part of a prompt.
About this term
- Level
- Beginner · Core ideas for first-time readers
- Category
- Basics
- Fields
- Language models
- Added
- 2026-10-07
How it works
The AI reads what is in the prompt and produces the most fitting answer to follow it. Besides the questions users write themselves, a 'system prompt' that the service puts in ahead of time may set the AI's tone or role. Even the same question can get very different answers depending on how it is phrased or what information is added.
Why it matters
Writing good prompts can improve the quality of results without changing the model. That is why designing prompts to get the desired answer is sometimes called 'prompt engineering.' Techniques such as assigning a role, providing examples and specifying the output format are widely used.
Things to know
There is a security problem called 'prompt injection,' in which instructions secretly inserted into a prompt make an AI behave in unintended ways. This needs particular attention with AI agents that read web pages or documents and take actions. Also keep in mind that personal information or company secrets placed in a prompt may be passed to an external service.
Example
Writing 'Summarize this text in three lines so that an elementary school student can understand it' instead of 'Summarize this text' gets you an answer much closer to what you want. Research covered on this site, such as LMBuild, also gives tasks through prompts and evaluates how well LLM agents carry them out.
Where it came up on this site
- briefing AWS offers Z.ai's 753B MoE model 'GLM 5.3' on Amazon Bedrock 2026-10-07
- Papers LMBuild: Evaluating LLM agents that build structures that can actually be assembled and work 2026-10-07