All activities Student learning
Students read six mock AI outputs through different perspectives, find who is stereotyped or missing, and rewrite the prompts to fix it.
Generative AI learns from huge amounts of text and images made by people, so it can repeat the patterns in that material, including stereotypes and blind spots. In this activity, teams read six printed mock chatbot responses to everyday prompts. Each team member reads through a different Perspective Lens, so the group notices more together than anyone would alone. Teams name the assumption in each output, identify who is missing, and rewrite the prompt or the output so it works for more people.
Hook
Give students 60 seconds to sketch "a scientist at work" on scrap paper. Hold several up. Ask: "What did most of us draw? Where did that picture in our heads come from?" Then say: "AI tools learn from what people have written and pictured, millions of times over. Today we check whether they picked up the same habits."
Facilitator noteKeep it nonjudgmental. The point is that everyone carries patterns, and AI learns patterns from us.
Explore
Each team member takes one Perspective Lens from Handout C and a highlighter color. Students read Handout A silently first, highlighting anything their lens makes them notice. Then they go round-robin: each person shares one thing nobody else highlighted. The Recorder tallies how many noticings came from only one lens.
Facilitator noteThe tally is your evidence for S3.3. Most teams find that several important noticings came from just one person.
Practice
Teams complete Handout B for at least four outputs: the assumption the AI made, the words that show it, who is left out or stereotyped, and a fix. Remind them to quote the output. Circulate and ask: "Is this output wrong, or just narrow? Would it hurt anyone if a student turned it in as-is?"
Facilitator noteSome outputs are subtle (Output 5 assumes home internet and a quiet room). If a team says "nothing is wrong here," ask them to read it through the lens they haven't used yet.
Create
Each team picks one output and uses the bottom of Handout C to write a better prompt (for example, adding "include characters of different genders and backgrounds") or a corrected output. If you have a district-approved chatbot, run two teams' revised prompts live on the projector and compare the results to the originals.
Facilitator noteA better prompt helps but doesn't guarantee a fair output. If the live result is still narrow, that's a great teaching moment: humans still have to check.
Debrief
Each team shares its sharpest noticing and its rewrite in 60 seconds. Then discuss: "Where does AI bias come from? Who should fix it: the company, the user, the teacher?" and "How did having different lenses on your team change what you found?" Close by asking students to finish this sentence aloud: "Before I use an AI answer about people, I will…"
Facilitator noteLook for students who connect bias to training data (patterns in what people have made) rather than saying the AI "has opinions."
Designed to run fully on paper. The six outputs are printed mock responses, clearly labeled as written for this lesson, so no device or account is needed. Colored highlighters make each lens's contribution visible on the page.
The teacher runs the original and revised prompts in a district-approved chatbot on the projector and pastes results into a shared class slide. Students compare live outputs with the printed mocks. Grades 9–12 students with district-approved accounts may test prompts in pairs and log differences. Students under 13 do not use AI tools directly.
Does it need a screen? A live run shows that real outputs vary each time and that a better prompt helps but doesn't guarantee fairness. That turns "AI can be biased" into evidence students observe. The printed mocks carry the core thinking.
What you should be able to see or collect if it worked.
Students read mock AI outputs through perspective lenses, quote words that reveal an assumption, and explain how training-data patterns produce stereotypes.
In the next research or writing assignment where AI is allowed, students add one line to their work: "I checked AI output for who's missing by…" Families can try the "draw a scientist" question at dinner and compare answers.
