The Algorithmic Arbiter Teacher Guide The Algorithmic Arbiter
Teacher Facilitation Guide • Grade 8 Social Studies
Duration 45 MIN
Learning Objectives
Differentiate between human-led "user flagging" and automated "algorithmic moderation."Identify the linguistic hurdles (sarcasm, irony, context) that cause AI to fail in moderation.Evaluate the societal risks of relying on non-transparent "Arbitrators of Truth."
Materials Needed
Slideshow
Sarcasm Test Worksheets
Red/Green Voting Cards
Scissors (to prep cards)
Pacing
0-5 MIN
Warm-up: Meme Voting
5-15 MIN
Video: Moderation Tools
15-35 MIN
The Sarcasm Test Activity
35-45 MIN
Reflection & Discussion
Procedure
1
Warm-up: The Moderation Vote (5 min)
Display the "Ambiguous Memes" on the slide. Have students use their Red/Green cards to vote if the post should be "SAFE" (Green) or "HARMFUL" (Red).
Teacher Note: Challenge students who disagree. Ask: "Why might someone find this offensive?" or "Is this a joke or a threat?"
2
Video & Concept Briefing (10 min)
Watch the video segment 04:29-06:09 . Focus on the definitions of User Flagging vs. Algorithms . Discuss the "High-Context" nature of language mentioned in the video.
3
Main Activity: The Sarcasm Test (20 min)
Step A: Pairs write 5 comments that are either sarcastic, ironic, or use slang. They must record their "True Intent" (Safe/Harmful) on their sheet.Step B: Pairs swap sheets. They now act as an "AI Bot." They are ONLY allowed to use a "Keyword Blacklist" (e.g., hate, kill, trash, weapon) to judge the post. They cannot use "common sense."Step C: Compare results. How many "Safe" posts were flagged as "Harmful" by the AI?
4
Reflection (10 min)
Discussion: "Why is it dangerous if an AI can't understand a joke?" Connect this to the term "Prior Constraint" from the video—if we block things before they even post, what do we lose?
Teacher Cheat Sheet: Key Vocabulary
Platform Law: Private company rules that act like government laws for users.
User Flagging: When humans report content (risk: "weaponized" flagging).
Section 230: Law protecting platforms from liability for user-posted content.
Prior Constraint: Blocking or suppressing content before it is even published.
The Algorithmic Arbiter Slides System Status: Online
The Algorithmic
Arbiter
Who decides what you see? Human judgment vs. AI Logic.
Civics in the Digital Age
// MISSION_OBJECTIVES
Compare Methods
Differentiate between human flagging and AI moderation.
Analyze Limitations
Explain why AI struggles with context, sarcasm, and irony.
The Flash Vote
Get your GREEN and RED cards ready.
Is the following post Safe or Harmful?
POST #001
@User_404
2 minutes ago
"I just LOVE sitting in gridlock traffic for two hours on a Monday morning. It's the highlight of my week."
SAFE
HARMFUL
POST #002
@StreetStyle
Just now
"My new sneakers are ABSOLUTELY KILLER. I'm going to destroy the competition at the race tomorrow!"
SAFE
HARMFUL
The Tech Behind the Scenes
Embedded media
Watch For:
How AI treats "high-context" linguistic techniques like sarcasm and irony.
Question:
Why is AI better at images than text?
The Sarcasm Test
Human vs. Machine: Who wins?
1
The Creators
Write 5 comments that use sarcasm, irony, or heavy slang.
Make sure some are "Safe" and some are "Harmful."
2
The AI Bot
You are a Strict Keyword Bot.
Ignore the "vibe." If you see a word on the "Blacklist," flag it. Otherwise, it's Safe.
Final Reflection
"Why is it dangerous if an AI
Consider the concept of Prior Constraint —censoring ideas before they are even shared.
The Sarcasm Test Worksheet The Sarcasm Test
Activity: Algorithmic Content Moderation
NAME:
DATE:
Phase 1: The Creators
In your pairs, write 5 social media comments . Your goal is to use sarcasm, irony, or slang to make the message's true meaning "hidden."Example: "Oh great, another math test. Just what I wanted for my birthday!" (True Intent: Annoyed/Safe)
| # | Your Comment (Use Sarcasm/Slang) | Your Intent
Phase 2: AI Bot Moderation
Swap Sheets with another group!
Keyword Blacklist
As the AI, if YOU see any of these words, you MUST flag the post as HARMFUL .
KILL WEAPON BOMB DESTROY FIGHT TRASH ATTACK STUPID
AI BOT JUDGMENT (Ignore the intended "vibe"):
| # | Trigger Word Found? | AI DECISION
Final Analysis
How many comments were mislabeled ? (AI thought it was Harmful but you meant Safe, or vice versa)
Based on your results, why do you think AI struggles with language moderation?
Moderation Voting Cards Moderation Voting Cards
Cut along the dotted lines. Use these cards to participate in the "Flash Vote" during the lesson warm-up.
SAFE
No Violation Detected
HARMFUL
Flag for Removal
How to Use:
When the teacher displays a post, decide if it violates community standards.
Hold up the GREEN card if you think it's harmless or context-dependent (like a joke).
Hold up the RED card if you think it's dangerous, misleading, or clearly hateful.
Be prepared to explain your choice to the class!
