This course remembers completed sections and your last course page in this browser on this device. RayAI does not receive this progress or anything you write. Unfinished answers and written responses are not saved, so an interrupted activity restarts when you return.
Module 1 | 25 minutes | Essentials Fast Track core + optional deeper learning.
Know what prompting can and cannot do
What you will learn
- Explain what a prompt can and cannot change.
- Recognize when AI is the wrong tool.
- Match human review to the possible consequence.
The essential idea
A prompt is the instruction and context you give an AI system.
Generative AI creates responses from learned patterns, the current context, and any tools it can use.
It does not automatically know your intent, your organization’s rules, the latest facts, or what could happen if it is wrong. A clearer prompt can make an answer more relevant and consistent, but it cannot guarantee that the answer is true.
When AI can help
- Draft, rewrite, summarize, organize, transform, or classify supplied material.
- Generate options, questions, explanations, outlines, and alternative perspectives.
- Compare information using criteria you define.
- Prepare a first pass for review.
A responsible person still reviews and improves the result.
See it in action
Weak prompt
Plan our community event.
Likely problem
Because the prompt gives no date, audience, venue, budget, activities, or success criteria, the model may invent them. A polished plan can still hide guesses.
Better start
Using the notes I provide, draft a planning outline for a community skills-sharing event. Separate confirmed facts from open questions. Do not invent dates, owners, costs, or commitments.
The more serious the consequence, the more review the answer needs. AI may draft or organize; a person remains responsible for important decisions and actions.
- AIproduces a draft
- Learnerreviews it
- Learnerchecks what matters
- Learnerdecides
Explore further
Why the better prompt works
The fix is not a magic phrase. Name the real task: draft from known facts, mark missing facts, and leave commitments to the organizer.
How much review does this task need?
Low
Example: Private brainstorming
AI role: Generate options
Human review: Preference and sense-check
Moderate
Example: Customer message or project plan
AI role: Draft and organize
Human review: Facts, commitments, tone, recipients
High
Example: Medical, legal, financial, safety, employment, or public claim
AI role: Summarize approved sources or prepare questions
Human review: Qualified review; AI is not final authority
Action-taking
Example: Send, publish, purchase, delete, change access, or move money
AI role: Propose or perform a bounded action
Human review: Preview, permissions, confirmation, logging, recovery
Practice deciding when AI is appropriate
For each task below, write three short answers.
- Is AI useful here?
- Who must make the final decision?
- What, if anything, should be checked before using the answer?
- Create five possible titles for a family photo album.
- Compare three insurance plans using the official plan documents.
- Rank job candidates from interview notes.
- Explain a contract clause in plain language.
Try it yourself
- Choose the suggested low-risk task or another simple one.
- Open an AI assistant, enter the prompt, and send it; do not include private or sensitive information.
- Read the answer and ask whether AI is appropriate and what a person must still check or decide.
- Return to the course and record the notes below.
If you cannot use an AI assistant now, read the worked example and write what you would try later.
Everyday life
Compare two ordinary options, then identify what information is missing before choosing.
Learning
Ask for an explanation of a familiar topic, then identify one claim you would verify.
Work
Draft a fictional meeting follow-up and name every commitment that requires human confirmation.
Your response stays in this page and is not saved or sent to RayAI.
Check your work
Quick reflection: What one change made the answer more useful or trustworthy? If you are unsure, compare your two prompts and name the clearest difference.
Knowledge check
Fast Track: answer Questions 1–3. Complete Course: answer Questions 1–4. After you attempt the questions for your path, review the answers and explanations.
These checks are for practice, not grading. Try each question before reviewing the explanation. Your choices and scores are not saved or sent to RayAI.
Complete Course reflection — apply the idea to your task
4. Think about your starting task. Who must make the final choice? What could go wrong if the AI answer is incorrect?
Your response stays in this page and is not saved or sent to RayAI.
A strong response names a specific person responsible for the final decision and a real consequence. Example: 'The project lead owns approval; a wrong deadline could create an unapproved commitment.'
Review the answers and explanations
Use these explanations to diagnose your thinking. A correct letter without the right reason is only partial learning.
Question 1
Best answer: A. A stronger prompt can improve the likelihood and consistency of a useful output. It does not guarantee truth or create professional authority.
Question 2
Best answer: B. Medication changes are high-stakes clinical decisions. AI may help prepare questions or summarize approved material, but a qualified clinician and patient must make the decision.
Question 3
Best answer: B. A customer message can create commitments or harm a relationship. Review facts, tone, policy, authority, recipient, and any promise before sending.
Question 4
A strong response names a specific person responsible for the final decision and a real consequence. Example: 'The project lead owns approval; a wrong deadline could create an unapproved commitment.'
Check your learning
Complete Course: 0 of 4 points recorded on this page.
Answer the questions for your selected path, then use the feedback to decide what to review.
Module takeaway
A better prompt can improve direction, but it cannot make AI certain or qualified. Use AI for a first pass, keep sensitive information out, check important claims, and let a person make decisions when consequences matter.
Optional deeper learning
Optional deeper learning — what can shape an AI response
A prompt can include text, examples, documents, images, earlier conversation, tool descriptions, and rules for the output.
An AI system may also follow higher-priority instructions you cannot see.
Optional deeper learning — coconut challenge: spot the highest-risk task
Coconut challenge: Which task could cause the most harm if the answer is wrong? In one sentence, explain why a confident-sounding AI answer would not be enough.
Finish this module
Mark this module complete when you have attempted every activity you intend to complete and reviewed your work against the v0.5 guidance.