Content Moderation
Domain: Safety & Ethics
🎯 Motivation
Automatically detecting and filtering harmful, offensive, or inappropriate content in AI outputs.
📋 Concrete Example
Moderation API: input ‘hate speech text’ -> output: {hate: 0.95, violence: 0.02, sexual: 0.01}. If hate > 0.7, block the content. Automated filtering at scale.
🔗 Analogy
A bouncer at a club - they check each person (content) and refuse entry to troublemakers (harmful content). Content moderation is the bouncer for AI outputs.