Learn with AI
A convincing AI answer can still be wrong
The short answer
A confident explanation can still be wrong. Check an important claim against trusted course material, inspect the reasoning and keep unresolved questions visible.
- Check the source itself.
- Inspect the connecting steps.
- Ask a tutor when sources disagree.
Understand
Confidence is not verification
Asking AI whether it is sure does not provide independent evidence. Neither does a citation that has not been opened.
Choose one important claim and find the passage or reasoning that supports it.
Check against a source

See the diagram
Check the claim, not the tone
See it in practice
Build a checked note
Compare the claim with a trusted course source. If it is wrong, correct the note and keep the checked explanation.
If you cannot resolve the difference, mark it as a question for the tutor. Do not turn uncertainty into a confident fact.
Explain it in your own words

See the diagram
Choose the important claim
Use AI thoughtfully
Where AI fits
Use AI responses as drafts to inspect. Accuracy on an earlier question does not guarantee accuracy now.
Make the next step clear

See the diagram
Inspect the reasoning
Try it today
Try one thing
Choose one important claim in today's AI explanation. Verify it using a course source and write the line of reasoning that supports it. If you cannot check it, mark it as unresolved.
Ask for a little help

See the diagram
Leave uncertainty visible
Check what changed
Can you explain why the answer is right without referring to the assistant's confidence? Can you point to the source or calculation that supports the conclusion?
Explore 8 more visual examples Optional. Pick the one you need.
Check against a source

See the diagram
Reassurance is not checking
Check against a source

See the diagram
Use your course material
Check against a source

See the diagram
Open the cited source
Ask for a little help

See the diagram
When sources disagree
Check against a source

See the diagram
Correct the study note
Check against a source

See the diagram
Do not generalise from success
Make the next step clear

See the diagram
Keep a short uncertainty list
Check against a source

See the diagram
No single error rate
What we know, and what we do not
Evidence and limits
NIST's document describes risks and risk-management considerations. It does not provide a single error rate for all models, subjects or prompts, and it does not test this classroom checklist. Verification remains necessary even when a tool has answered similar questions correctly before.
