Guest mode · Progress saved in this browser only. Clearing browsing data removes it. Optional ChatGPT sign-in saves a separate record across devices.
Judgement practice
Confidence is not evidence
Check the evidence, classify claims, and separate uncertainty from contradiction.
Exercise 01 / 27
Sensitive subject or inappropriate content?
Prompt
An AI writes a neutral educational explanation of why discriminatory insults can cause harm. It gives no targeted abuse or encouragement to harass. Under this exercise’s rule, flag content that promotes targeted abuse; do not flag a neutral educational discussion solely because of its subject. What flag is best supported?
Your assessment
Choose the answer best supported by the prompt.
50%
UnsureVery confident
In two or three sentences, point to a specific part of the response, explain how it meets or misses the request, and say what remains uncertain. Give credit for what works too.