Beginner level · Module 5: Responsible AI at Work · Lesson 2 of 3
Fact-Checking AI Output
Build a quick, reliable habit of checking AI outputs before they reach customers.
In this lesson you will
- Apply a fact-checking routine
- Decide which outputs need extra review
Key concepts
facts, figures and namestrusted sourcessecond-person reviewrisk level
The lesson
Before sending AI-drafted text, check every price, date, name, phone number, quantity and promise against your own records.
Match the check to the risk: a social media post needs a careful read; a contract summary or health information needs checking against the original and a second person.
If something cannot be verified, remove it rather than risk misleading a customer.
Step by step
- Check an AI-drafted message line by line.
- Correct any errors found.
- Classify five outputs by risk level.
- Write the checking rules for the guide.
Practice activity
Write the accuracy section of the staff AI guide and check all project outputs.
Expected result: Checking rules and corrected project outputs.
Check your understanding
-
What must be checked in a customer message drafted by AI?
Show answer
Prices, dates, names, numbers and promises.
-
What should you do with a claim you cannot verify?
Show answer
Remove it.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.