Practical Prompt Engineering – Reliable Workflows for Human Users
Lesson 2: Settings, Repeatability, and Why Answers Change
Lesson Objectives
By the end of this lesson, students should be able to:
- Explain why the same prompt may produce different answers.
- Choose lower-randomness behavior for factual or structured work.
- Use output length limits to control cost and focus.
Lesson Content
AI answers can change because language models choose from many possible next words. Some tools expose controls for randomness, length, and repetition. Other tools hide those controls but still make similar decisions behind the screen.
For student work, the practical rule is simple: use more controlled behavior for factual, legal, technical, or structured tasks. Use more flexible behavior when brainstorming names, story ideas, ad angles, or creative variations.
Length also matters. If you ask for "everything," the AI may produce a long answer that is hard to review. A better approach is to request a specific format, such as a five-row table, a checklist, or three options with pros and cons.
The setting is not the lesson by itself. The real skill is knowing what kind of result you need. A customer policy needs consistency. A logo tagline brainstorm may benefit from variety.
Enroll to continue this lesson.
The preview above shows the lesson objectives and opening lesson content. Enroll to view the full lesson, complete the practice work, and take the lesson quiz.