Background readingEstablished
Prompt Engineering
noun
Designing and testing the instructions, examples and context given to an AI so its output better fits a particular task.
A little background for the cultural references. This is a supporting explanation, rather than a slang or meme entry. Explore the AI culture dictionary.
What it means: Prompt Engineering
Prompt engineering is the deliberate work of specifying what a model should do and evaluating how well the instructions work. That can include examples, an output format, relevant background and constraints. It is not a guarantee of correct answers, and it does not require a secret phrase or a grandiose role description.
How people use it
The useful bit of prompt engineering was giving it a good example of the report, not telling it to be a genius.
The label can be serious or ironic. In a technical workflow, it implies testing against the task. In casual conversation, it may simply mean being unusually particular about how you ask an AI for something.
Illustrative example written for this dictionary.
Why it matters
It highlights a basic change in software interaction: ordinary language becomes part of the interface. Good instructions can reduce ambiguity, but the responsibility to inspect results remains with the person or system using them.
For another angle, explore Promptmaxxing and Vibe Coding.
Where it came from
Prompting practices developed alongside language models and became widely visible as conversational AI reached a broader public. The cultural phrase expanded to cover everything from a professional evaluation workflow to someone rewriting a request until a chatbot cooperates.