Prompt Engineering
Structuring instructions and context to reliably get useful output from a model.
CurrentbeginnerBeginner-friendlyFull course available
Overview
Prompt engineering is the practice of structuring instructions, examples, and context to reliably steer an LLM's output -- specifying format, providing examples (few-shot prompting), and being explicit about constraints, since models respond to exactly what's written, not what's intended.
- What it is
- The practice of designing instructions and context to reliably shape a language model's output.
- Why it's used
- The same underlying model can produce dramatically different quality output depending on how a task is specified.
- Where it fits
- Applies to any application built on an LLM -- this platform's own AI tutor prompt (when configured) is a real, inspectable example.
Core concepts
- Clear, specific instructions
- Few-shot examples
- Structured output formats
- Separating instructions from untrusted data
Example
Specificity about format and constraints (length, structure, focus) is usually the single biggest lever for improving output quality -- more than clever wording.
// Vague: "Summarize this."
// Specific: "Summarize this in 3 bullet points,
// each under 15 words, focusing on action items only."Common use cases
- Building reliable AI-powered features
- Getting consistent, structured output from a model
Project ideas
- Take a vague instruction and iteratively rewrite it three times, making it more specific and structured each time