Activity

Same Prompt, Same Answer? Comparing AI Outputs

Grades 9-12

Subjects: Artificial Intelligence, Communication & Collaboration, Critical Inquiry & Meaning Making, English Language Arts

Overview

Students enter the same prompt into two or three AI tools and compare the outputs as texts. They code shared and unique ideas, examine structure and language, and identify what none of the outputs included. The central idea: agreement among AI tools is not verification. Tools built on overlapping data tend to share the same gaps, so matching answers may carry no more weight than a single one.

Career connection: Journalists, fact-checkers, researchers, and policy analysts check whether sources are truly independent before treating agreement as confirmation. This activity builds the same habit by asking where information comes from, not just how often it is repeated.

NB Curricular Connections

English Language Arts 9, 10, 11 and 12

Strand: Reading – Big Idea: Text Analysis/Criticality – Skill Descriptor: Critically analyze and evaluate the characteristics, language, form, features, and genre of a range of texts.

What You’ll Need

Student devices with internet access
Students’ school Copilot accounts
One or two free AI tools that do not require an account (for example, ChatGPT or EcoGPT)
Teacher-generated sample outputs from two or three tools, printed or shared digitally (for students who do not use AI)
Two highlighter colours (paper or digital)
Prompt for activity: “Act as an experienced literary critic. Write a concise summary (150-200 words) of the main themes in [title of novel, short story, or poem] for [audience]. Use clear paragraph structure and plain language.”
A simple comparison chart: Idea | Tool A | Tool B | Tool C

Instructions

Before class: For students who do not use AI, run the prompt through two or three tools and save the outputs as samples. Students who prefer not to use AI, or who cannot access the tools, complete the activity using these samples.
Set up tools: Students use their school Copilot account plus one or two free tools that do not require a login, such as ChatGPT or EcoGPT. Gemini or Claude may be used only if a student already has an account and is comfortable using it for this activity. Do not encourage students to create a new account for this activity.
Open with a question: If several AI tools give the same answer, does that make the answer true?
Generate outputs: Students enter the prompt in each tool, substituting novel/story and audience for a relevant one. Remind students not to enter personal information or additional information beyond the existing prompt.
Code the outputs: Students highlight ideas that appear in every output in one colour and ideas that appear in only one output in another. Record them in the comparison chart.
Analyze form and language: Compare the order of ideas, tone, word choice, and structure. Discuss what genre conventions the outputs share (for example, opening with benefits and closing with a balanced conclusion).
Read for absence: List what none of the outputs mentioned. Consider whose perspectives are missing, such as local, Francophone, or Wabanaki viewpoints, or anything specific to your school or community.
Check against a human source: Compare the outputs with a credible article, a librarian, a teacher, or a family or community member.

Assessment

Discuss: Why might different tools produce such similar text? Connect the answer to shared training data and common sources.

Write: Students write the paragraph none of the tools could have written, drawing on their own context, experience, or the human source they consulted.

Make Meaning

Please see the attached PDF for several choices on how you and your learners can reflect upon today’s activity

Apply

Grade 9: Students explain how purpose, audience, and genre shaped the outputs, then choose one shared claim and evaluate whether its reasoning is sound or fallacious.
Grade 10: Students challenge one idea that every tool repeated, using their “read for absence” list to explain why the shared answer is incomplete.
Grade 11: Students revise one output to include a perspective missing from all of them, drawing on a human or published source rather than their own assumptions.
Grade 12: Students write a reflection on how the activity shifted their view of AI agreement and how they will check information in the future.
All grades: Enter the same prompt into one tool three times and compare the results. Discuss whether differences between tools are meaningful or just variation between runs.

Global Competencies (optional)