In the first exercise, AI used the information you provided to make a suggestion that sounded confident and reasonable. An answer can sound just as confident and still be wrong. You can spot this by comparing it with the input.
A language model continues text
A language model writes an assistant's response one piece at a time, based on what text is likely. A likely continuation can include information your material does not support; this is called a hallucination. It can sound as confident as correct information, so confidence alone does not tell you whether it is true.
Input, suggestion and decision
Throughout this course, you will read each answer using the same distinction between input, suggestion and decision as in the first exercise. This helps you know what to check:
- When you provide your own material, check whether the suggestion stayed within it.
- When you ask for current information, check which sources support the suggestion.
- When you compare answers or make a choice, the decision and its reasons are yours.
How to spot an addition
Pick a specific detail from the answer, such as a number, name or time, and look for it in the input. If it is not there, AI added it.
Free practice on this website
Find what came from the input and what was added
The material contains a fictional request and an AI response. Answer the questions using the material.
Your answers stay only in this browser tab. They are not sent to the AI Licenses server or analytics.
Question 1/2
Question 1/2
List two details in the response that are not in the request.
Skills to practice
Transform material faithfully
Sources and review
- Reviewed:
- September 11, 2026