Prompt
This is the final retrieval check. Answer from memory — you're about to integrate everything you've learned.
- Name the four reliability zones and give a task example for each.
- What are the four elements of an effective prompt?
- Name the four failure categories and explain how to document a failure you find.
- What sensitivity level applies to salary data? What service tier is appropriate?
- When would you use a research engine (like Perplexity) instead of a general chatbot (like ChatGPT)?
Answers
- High (structured text from your notes — e.g., formatting meeting notes into a summary), Moderate (creative/analytical text — e.g., drafting a client proposal), Low (requires specific facts you can't verify — e.g., competitive analysis with market stats), Not Suitable (legal, medical, financial with regulatory consequences — e.g., drafting a contract clause).
- Role (who you are), Context (what you're working on and for whom), Format (structure, length, tone), Constraints (must-haves, boundaries, specific details).
- Hallucination/factual inaccuracy (AI invents facts), tone mismatch (wrong voice for audience), missing constraint (left out something you asked for), format error (wrong structure or length). Document each with: (a) quote the specific failing text, (b) name the category, (c) explain why it matters for this task.
- Salary data is Restricted — personally identifiable compensation information. Do not use with general AI tools. Enterprise tools with BAA may be acceptable with IT/legal approval.
- Use a research engine when you need facts with citations you can verify. Chatbots generate plausible text; research engines search the web and cite sources. If the task requires you to know something is TRUE (not just well-written), use a research tool.
Bridge
If you answered all five confidently, you're ready for the full integration exercise. If you struggled with 2+, go back to the relevant module: reliability spectrum (Module 2), prompt elements (Module 3), failure categories (Module 3), data sensitivity (Module 4), or tool selection (Module 4). Then return here.