Teach Students to Fact-Check AI: A Classroom Lesson on Reverse Searching
Artificial intelligence is no longer something students encounter only in STEM clubs or computer science electives, it’s part of everyday learning. That reality means we need more than digital citizenship lessons: we need AI literacy.
One of the most compelling ideas to support this comes from Halaweh’s 2023 article ChatGPT in Education: Strategies for Responsible Implementation, which introduces a skill called reverse searching. As the author explains:
“Reverse searching is a new concept… through which students use outputs to find supporting evidence and references for text generated by ChatGPT” (Halaweh, 2023, p. 4). EJ1385551
This approach reframes AI from a source of answers to a starting point for inquiry (something my science educator side of me loves), and it’s exactly the kind of shift our classrooms need.
Below is a complete sample lesson for grades 7-12 to help students develop AI literacy
Don’t Take AI’s Word for It
Grade Level: 7–12
Time: 1–2 class periods
Goal: Students learn to verify AI-generated information using the reverse searching method described by Halaweh (2023).
Skills: critical thinking, research literacy, AI literacy, source evaluation, paraphrasing, evidence tracking
Objectives
Students will be able to:
Prompt an AI tool (ChatGPT, etc.) to generate an explanation or claim.
Identify statements within the AI-generated text that need verification.
Use reverse searching to confirm or challenge AI claims using credible sources.
Document their verification process through citations and a mini-reflection.
Explain why human-AI collaboration improves accuracy and academic honesty.
Warm Up
Start class with a discussion:
“Is AI always correct?”
“If it sounds confident, does that make it accurate?”
“What is our responsibility when we use information we didn’t generate ourselves?”
Introduce the article’s central idea:
Explain that reverse searching is the process of verifying claims after AI presents them, just like fact-checking a friend who says something suspiciously confident.
Step 1: Generate AI Output
Students give ChatGPT a prompt like:
“Explain why the Great Barrier Reef is threatened.”
“Summarize how photosynthesis works.”
“Describe the causes of the French Revolution.”
“Explain how vaccines work.”
Choose a topic already in their curriculum, or use a topic relevant to another class all students share (social studies is always a good place to start and can have students verify with primary source documents). Be careful with choosing complicated science prompts with younger students as many of the academic articles can be beyond their capabilities.
Step 2: Identify Claims That Need Verification
Students highlight:
Factual claims
Definitions
Statistics
Dates
Cause/effect explanations
They label each one with: Check, Double-Check, or Seems Unclear.
Step 3: Reverse Search
For each bolded claim, students:
Search credible databases (Newsela, Britannica, .gov, .edu sources). This will be the most intensive part of the project as you will need to guide students through the school’s resources and how to access and search.
Verify whether the AI-generated claim is Accurate, Partially Accurate, or Inaccurate.
Record the evidence and cite the source.
Students quickly see that:
AI sometimes hallucinated references
Dates and statistics may be off
Explanations may be oversimplified or missing nuance
This is exactly the critical thinking Halaweh emphasizes when arguing that:
“Students should document… contradictory findings, texts without references, new ideas or developments, [and] any judgments made by the student that ChatGPT did not support” (Halaweh, 2023, p. 6). EJ1385551
Step 4: Rewrite the AI Content Using Verified Evidence
Students reconstruct a corrected version of the paragraph using:
Only verified information
Their own words
Proper citations
This enforces both academic honesty and deeper comprehension.
Step 5: Reflection (3–5 sentences)
Students answer:
What did AI get right?
What needed correction and why?
How did reverse searching make you think differently about AI-generated text?
When should people not use AI without verification?
Remind students of Halaweh’s warning that:
“Texts generated by ChatGPT copied and submitted as your final writing is considered plagiarism.” (p. 6). EJ1385551
This reinforces both ethical use and skillful use.
This is an excellent place to teach students how AI is trained and that the technology is only as good as the information that is put into it.
Extension: The Accuracy Score Challenge
This is my student’s favorite part of the lesson. In groups, students compete to generate an input in which the output has the most hallucinations or errors. The creativity and fervor with which your students fact check will be both humorous and impressive.
Conclusion
Reverse searching flips the script on AI use. Instead of asking whether AI will harm learning, we give students structured opportunities to question, verify, evaluate, and think independently.
As the article argues:
“By combining ChatGPT and human authors, the output is superior in terms of creativity, originality, and efficiency” (Halaweh, 2023, p. 4). EJ1385551
This lesson doesn’t just allow AI in the classroom, it transforms AI into a tool that strengthens critical literacy and academic integrity.
