Bias, stereotypes, and unfair assumptions
Unit ID: AIT-M11-U02 Estimated active time: 15-20 minutes
The main idea
AI output can repeat unfair patterns or invent reasons about people.
It has learned from text written by people, including the parts that carry assumptions nobody would defend if stated plainly.
Why this matters
The output is fluent and neutral-sounding, which makes an unfair assumption easier to accept than it would be from a colleague.
Safe example
A tool guesses why a person missed a deadline without evidence.
The guess is plausible, invented, and about someone who is not in the conversation.
Better working habit
Remove unsupported judgements and ask for neutral wording.
Replace the guess with the fact: the deadline was missed. Why is a question, not a sentence.
What to watch for
Do not use AI to infer character, ability, honesty, or intent.
These are not knowable from the text, and an inference recorded once tends to be repeated by people who never saw the evidence.
Mini practice
Rewrite biased or assumptive sentences neutrally.
Then run the swap test: change the name or background and see whether the wording still feels right.
Check yourself
Before moving on, answer:
- Is any statement here about someone's motives or character?
- What is the evidence for it?
- Would this read differently with a different name attached?