4 · Verify the output (hallucinations)

When an AI states something false as if it were true, that's called a hallucination, and it happens often, because the model is generating plausible text, not checking facts. Hallucinations are dangerous precisely because they are confident and well-written.

Build a verify habit:

• Treat every fact, quote, statistic, name, and citation as unconfirmed until you check it.
• Ask for sources, then open them. Models can invent real-looking citations (fake DOIs, fake URLs, fake court cases). A source you didn't open is not a source.
• Cross-check important claims against a primary or reputable source.
• Be extra skeptical of recent events, numbers, legal/medical/financial advice, and anything you'd act on.

This is the heart of Learn.WitUS: every claim ties to a source you can verify. Use AI to draft and explain, and use real sources to confirm. "Trust, but verify" becomes "draft with AI, verify with sources."

Sources

• National Institute of Standards and Technology. (2023). AI Risk Management Framework (AI RMF 1.0), see "Valid and Reliable." https://www.nist.gov/itl/ai-risk-management-framework