Professional learning · Teachers, Librarians, Coaches, Leaders · 45 minutes
Explain It to Grandma
Educators draft 90-second, jargon-free explanations of AI concepts and test them live on colleagues playing a skeptical grandparent, a second grader, or a school board trustee.
Families, students, and board members are forming opinions about AI in schools right now, often from headlines. The educators who can explain it plainly and honestly, without hype or fear, shape that conversation. In this session participants pair an audience card with a concept card, draft a 90-second explanation, and then test it on a colleague who plays that audience and pushes back with the audience's real questions. Explanations get scored for plain language and for accuracy, because simple is only useful if it's also true.
In this packet
- Handout A: Audience and Concept Cards (cut-apart cards)
- Handout B: 90-Second Explanation Planner (worksheet)
- Handout C: Plain and Accurate Rubric (rubric)
- Facilitator key (last page)
TCEA ELE indicators
- AI1.2 I can explain AI concepts in simple terms to various stakeholders.
- AI1.1 I know how AI systems work and their capabilities in educational contexts.
- AI1.3 I am aware of the current limitations and potential biases in AI systems.
Full facilitator guide: mglearn.github.io/eles/activities/explain-it-to-grandma.html
Handout A for Explain It to Grandma
Audience and Concept Cards
Cut apart. Keep Audience (A) and Concept (C) cards in separate face-down piles. Concept card backs are the accuracy key; print them on a separate sheet.
Sheet 1 of 2
A1 · Audience
GRANDPARENT, 74. Raised three kids, uses a smartphone for photos and video calls. Pushback: "So is it a robot? Does it know things like a person does?" and "Why would you let a child use something that makes things up?"
A2 · Audience
SECOND GRADER, 7. Loves dinosaurs, asks "why" five times in a row. Pushback: "Is it alive?" "Does it have feelings?" "Why did it say a T. rex had feathers AND didn't?"
A3 · Audience
SCHOOL BOARD TRUSTEE. Busy, practical, answers to voters. Pushback: "What problem does this solve for our students?" and "What happens when it goes wrong, and who is responsible?"
A4 · Audience
HIGH SCHOOL SENIOR. Uses AI apps daily, thinks they know more than you. Pushback: "Everybody uses it. Why does it matter how it works?" and "If it's usually right, why check?"
A5 · Audience
LOCAL NEWS REPORTER. Looking for a clear quote. Pushback: "In one sentence, is AI good or bad for kids?" and "Are teachers using it to grade?"
A6 · Audience
PARENT WORRIED ABOUT PRIVACY. Read a scary article last week. Pushback: "Where does what my kid types go?" and "Can it be used to profile my child?"
A7 · Audience
PARENT WHO SPEAKS SPANISH, WITH A SCHOOL INTERPRETER. Warm and interested, new to the district. Pushback (through the interpreter): "Will this help my child learn English, or do the work for her?"
A8 · Audience
CAMPUS STAFF MEMBER (front office). Keeps the school running and hears everything from families. Pushback: "What should I tell parents who call about AI?" and "Should I be using it for emails?"
Handout A for Explain It to Grandma
Audience and Concept Cards
Cut apart. Keep Audience (A) and Concept (C) cards in separate face-down piles. Concept card backs are the accuracy key; print them on a separate sheet.
Sheet 2 of 2
C1 · Concept
What is a generative AI chatbot, and how does it write an answer?
C2 · Concept
Why does AI sometimes make things up ("hallucinate")?
C3 · Concept
How can AI be biased?
C4 · Concept
What happens to the information students type into an AI tool?
C5 · Concept
Why might two students get different answers to the same question?
C6 · Concept
Can the school tell if a student used AI to write something?
C7 · Concept
Will AI replace teachers?
Handout B for Explain It to Grandma
90-Second Explanation Planner
Write for the person on your Audience card, not for yourself. Aim for 90 seconds spoken aloud.
My audience and my concept. The ONE sentence I want them to remember:
An everyday comparison they'd recognize. Where does my comparison break down?
One honest limitation or risk I will name:
One thing my audience can do (a question to ask, a habit, a place to learn more):
Jargon swap list: the technical words I'll avoid and what I'll say instead.
After role-play: what I changed, and why.
Handout C for Explain It to Grandma
Plain and Accurate Rubric
Checker: score plain language and accuracy separately. A simple explanation that's untrue is not a win.
| Criterion | Beginning | Developing | Proficient | Exemplary |
|---|---|---|---|---|
| Plain language | Relies on jargon; audience would be lost. | Some jargon explained, some left in. | Everyday words throughout; one clear comparison. | Vivid, audience-specific comparison; no jargon; easy to repeat to someone else. |
| Technical accuracy | Contains a false claim (e.g., "it searches the internet for every answer," "it thinks"). | Mostly accurate, with one misleading simplification. | Accurate at the level of detail given; matches the concept key. | Accurate and names where the simplification stops being true. |
| Honest about limits | All hype or all fear. | Mentions a limit vaguely ("it's not perfect"). | Names one specific limitation or risk. | Names a specific limit and why it happens, without alarm. |
| Actionable | Audience left with nothing to do. | A general suggestion ("be careful"). | One concrete action or question for the audience. | A concrete action tailored to this audience's pushback question. |
Facilitator only for Explain It to Grandma
Answer key and notes
Handout A: Audience and Concept Cards
- A1 · Audience
- A2 · Audience
- A3 · Audience
- A4 · Audience
- A5 · Audience
- A6 · Audience
- A7 · Audience
- A8 · Audience
- C1 · Concept
- Accuracy key: it generates text by repeatedly predicting a likely next piece of text (a token), based on patterns learned from very large amounts of training text. It doesn't look answers up the way a search engine does, unless the tool is also connected to search or documents. Not accurate: "it thinks," "it understands like a person," "it searches the internet for every answer."
- C2 · Concept
- Accuracy key: because it produces likely-sounding text, not verified facts. When it lacks a strong pattern for a specific fact, it can generate a plausible but false name, date, number, or citation, in the same confident tone. Remedy: verify important claims with trusted sources.
- C3 · Concept
- Accuracy key: models learn from human-made data, which contains gaps, stereotypes, and over- or under-representation of some groups and languages. Outputs can reflect those patterns. Bias can also come from design choices and how a tool is used. Remedy: compare outputs, question defaults, and keep humans deciding.
- C4 · Concept
- Accuracy key: it depends on the tool's terms and settings. Some tools store conversations or may use them to improve the product; others, often those under district agreements, limit that. Families should know which tools the district has reviewed, and students should never enter personal information into any AI tool.
- C5 · Concept
- Accuracy key: chatbots usually add controlled randomness when choosing each next word, so the same prompt can produce different wording or even different claims. Tools, versions, and earlier messages in the conversation also change answers. Different answers are a reason to check, not a sign someone cheated.
- C6 · Concept
- Accuracy key: not reliably from the text alone. AI-writing detectors are known to be unreliable, including false positives, and concerns have been raised that they flag non-native English writers unfairly. Better evidence comes from the learning process: drafts, notes, conversations, and in-class work.
- C7 · Concept
- Accuracy key: honest answer, no one knows the future, but AI can't do the relational, judgment-heavy parts of teaching: knowing a child, noticing when something's wrong, deciding what's worth learning. It can help with some tasks, and teachers decide where it belongs, if it belongs at all.