A source you cannot open
The answer names a study, quote, court case, or article but gives no working path to the original.
A practical AI literacy course
Think clearly with AI in the room.
Learn what generative AI is actually doing, where its answers can fail, and how to make better decisions before you trust, use, or share an output.
Lesson 1 · How it works
AI predicts. It does not know.
A generative AI system creates a response by predicting likely pieces of text, images, or sound from patterns in its training data. It can produce a fluent answer without understanding the world the way a person does.
1You give it a prompt
Your words provide instructions and context.
2It matches patterns
The model uses statistical relationships learned from data.
3It predicts a response
The output is assembled one likely piece at a time.
4You verify the result
A person still has to check truth, context, and consequences.
Key idea
Confidence is a writing style, not evidence. A polished answer can still be false, incomplete, biased, or invented.
Lesson 2 · Warning signs
Know when to slow down.
These clues do not automatically make an answer wrong. They mean the answer needs stronger verification before you rely on it.
Suspiciously exact details
Precise dates, statistics, names, and quotations can be fabricated. Specificity is not the same as accuracy.
Only one point of view
The response treats a contested question as settled or leaves out the people most affected by the decision.
High stakes, easy certainty
Health, legal, financial, safety, and crisis advice needs qualified human judgment and current authoritative sources.
Lesson 3 · A repeatable method
Use the HUMAN check.
Run through these five questions whenever an AI output could affect your work, your choices, or another person.
- H
Hold before acting
How serious would it be if this answer were wrong? The higher the stakes, the more verification you need.
- U
Understand the source
Can you trace the important claims to reliable, current, original evidence outside the AI response?
- M
Minimize sensitive data
Have you removed private, identifying, confidential, or copyrighted material that you do not have permission to share?
- A
Ask what is missing
Whose perspective, context, or lived experience might the answer overlook? Who could be harmed?
- N
Name your responsibility
Are you prepared to own the final choice, rewrite the work in your voice, and disclose AI use when required?
Lesson 4 · Match effort to risk
Not every use needs the same response.
Low stakes
Brainstorming and formatting
Examples: title ideas, practice questions, reorganizing notes.Review quickly and keep your own voice.Medium stakes
School and public information
Examples: research summaries, presentations, recommendations.Open the sources, compare claims, and follow disclosure rules.High stakes
Health, safety, money, and rights
Examples: symptoms, contracts, emergencies, major financial choices.Do not rely on AI alone. Use qualified people and authoritative sources.Practice · Choose your response
What would you do next?
Select an answer and get immediate feedback. You can change your choice and try again.
Scenario 1
An AI gives you a surprising statistic and names a research paper, but provides no link. What should you do first?
Scenario 2
A friend asks you to paste a private group chat into AI so it can write a response. What is the best choice?
Scenario 3
AI writes a polished paragraph for a class assignment. What keeps the work genuinely yours?
Take it with you
Hold. Understand. Minimize. Ask. Name.
The HUMAN check will not answer every ethical question. It gives you a reliable place to begin before an AI output becomes a human consequence.