Analyzes AI/LLM applications for algorithmic bias, fairness violations, lack of diversity/representation, accessibility issues, transparency gaps, and explainability concerns. Focuses on discrimination across protected classes (race, gender, age, disability), fairness metrics, inclusive design, and ethical AI principles. Complements ai-safety (harm), ai-security (exploits), and ai-privacy (data rights). Triggers when reviewing hiring/lending/scoring systems, training data, model outputs, accessibility, or when user asks about bias, fairness, or ethics.
This skill identifies ethical issues and fairness violations in AI systems. It focuses on ALGORITHMIC BIAS, DISCRIMINATION, ACCESSIBILITY, and TRANSPARENCY - not safety harm, security exploits, or privacy violations.
AI Ethics (this skill):
AI Safety (different skill):
AI Security (different skill):
AI Privacy (different skill):
Under anti-discrimination law, protected classes include:
Types of Bias:
Training Data Bias:
Algorithmic Bias:
Deployment Bias:
RED FLAGS:
PROTECTED CONTEXTS:
FAIRNESS METRICS:
REGULATIONS:
MITIGATIONS:
RED FLAGS:
REQUIREMENTS:
MITIGATIONS:
Web Content Accessibility Guidelines (WCAG) 2.1/2.2:
Level A (Minimum):
Level AA (Required for compliance):
Level AAA (Enhanced):
RED FLAGS - AI-Specific:
REGULATIONS:
REQUIREMENTS:
MITIGATIONS:
The Black Box Problem:
RED FLAGS:
REGULATIONS:
REQUIREMENTS:
MITIGATIONS:
Principles:
RED FLAGS:
MITIGATIONS:
Carbon Footprint Concerns:
RED FLAGS:
MITIGATIONS:
Digital Divide:
Economic Displacement:
Cultural Sensitivity:
CONSIDERATIONS:
CRITICAL - Clear discrimination, illegal bias
HIGH - Significant fairness issues
MEDIUM - Ethical gaps, should improve
LOW - Best practices, enhance ethics
Privacy is a fundamental right. Respect it.