Prompt engineering
In short: Prompt engineering is the art of phrasing instructions to a language model so that the answer is good and usable.
In more detail: Good prompts state task, context, desired format and examples. The clearer the goal, the better the result.
In Depth
- Role and goal: “You are an editor. Shorten the text to 100 words.”
- Provide context: audience, tone, constraints.
- Examples (few-shot): Two or three examples show the desired pattern.
- Step by step: For calculations or reasoning tasks, ask for a step-by-step approach.
- Fix the format: list, table, JSON.
- Check: Read the result critically, do not accept blindly.
- Sensitive data: Do not paste passwords or personal data into cloud services.
See also: Hallucination (AI), Temperature and sampling