Unit 1, Day 2: Confident Isn't Correct

Lesson 2 of 6 in Unit 1, Use It, Check It, Own It

School-year day

Day 2 of 180

Time

20-minute core + 30-minute extension (50 minutes in all)

Teacher Talk with MystrS. Checked October 25, 2026 (Pacific time). Tags: Cyber, AI.

Existing article: /ai/check-an-ai-answer/

Goal: Students will be able to:

• Explain that a chatbot can say wrong things in a confident tone.
• Use a three-step check: find the claim, find a real source, compare.
• Decide whether an AI answer is safe to use, needs fixing, or should be tossed.

Materials:

• Projector and a teacher-run chatbot
• Two or three AI answers on a topic you teach, made ahead of time, with at least one error students can check
• One trusted reference: a textbook, an encyclopedia, or a government site

Core (20 min)

1. Show an AI answer that sounds great and contains a mistake. Students rate their trust from 1 to 5. (3 min)
2. Teach that sounding sure is not the same as being right. The National Institute of Standards and Technology calls made-up AI content "confabulation", and OpenAI's own guidance says "Confidence isn't reliability." (Both are cited in the existing article.) (5 min)
3. As a class, highlight the claims in one answer and check one of them against the trusted reference. (9 min)
4. Check for understanding. (3 min)

Extension (up to 30 min)

• Each group takes a different AI answer, highlights every factual claim, and checks two. (20 min)
• Groups report safe, fix, or toss, and explain why. (10 min)

Check for understanding: "Name one claim from today's AI answer that you checked. What source did you use, and what did you find?"

Standards (from our standards map, key: check-an-ai-answer):

• CSTA 2026: HS-ALG-PS-05; MS-ALG-PS-05