“A 2025 study found that handwritten notes improve recall by exactly 42%.”
No author, journal, link, or study title is provided.
⋮⋮ Drag case AI LITERACY LAB · CASE 08
The AI Evidence Lab
AI can sound certain even when its evidence is not. Examine each case, then decide whether the content is supported, needs verification, or raises a red flag.
MissionClassify all 8 cases with at least 80% accuracy.
Incoming cases8 remaining
“Stop taking prescribed medication for 48 hours before trying this supplement.”
The response gives individualized medical direction without a clinician or source.
⋮⋮ Drag case“A $50 item discounted by 20% costs $40.”
The response shows: $50 × 0.20 = $10; $50 − $10 = $40.
⋮⋮ Drag case“The claim appears in Rivera (2024), Journal of Digital Learning, doi:10.9999/jdl.2024.118.”
The DOI does not resolve and the journal cannot be located.
⋮⋮ Drag case“The company’s current CEO is Maya Chen.”
The answer may have been generated from older training data.
⋮⋮ Drag case“Alt text is a written description that communicates an image’s purpose or meaning.”
This matches guidance on the official W3C Web Accessibility Initiative site.
⋮⋮ Drag case“This contract clause is illegal in every U.S. state.”
No jurisdiction, date, statute, or legal source is identified.
⋮⋮ Drag case“The museum opened in 1927.”
The AI gives no citation, but the museum has an official history page.
⋮⋮ Drag casei
Drag a case to the best evidence zone.
On a touch screen or keyboard, select a case first and then select a destination.
DESIGN BRIEF
Assignment alignment
Learning Objective
After completing the activity, learners will classify eight AI-generated statements by evidence quality and appropriate next action with at least 80% accuracy.
Target Audience
College students and early-career professionals who use generative AI for research or workplace tasks and have basic AI familiarity but no formal fact-checking training.
Drag-and-Drop Mechanic
Learners examine eight case cards and drag each one into Supported, Verify, or Red Flag. Each card contains an AI claim plus an evidence clue.
Feedback System
Correct placements lock into the selected zone and receive case-specific reasoning. Incorrect placements return for another attempt with a targeted hint. Progress and accuracy update immediately.
Connection to Learning
Sorting requires learners to evaluate the quality and risk of evidence instead of merely recognizing a memorized definition. Immediate, case-specific feedback corrects misconceptions at the moment of judgment, while repeated classification builds a practical verification habit that transfers to real AI use.