Why AI Ethics Matters
AI systems make consequential decisions: loan approvals, resume screening, fraud detection, medical diagnosis. Unlike traditional software, AI decisions are often opaque, hard to contest, and systematically biased.
| Case | System | Harm | Root Cause |
|---|---|---|---|
| COMPAS Recidivism | US court risk scoring | Falsely flagged Black defendants as high-risk | Training data reflected historical racial bias |
| Amazon Hiring Tool | Resume screening AI | Discriminated against women | Trained on 10 years of male-dominated hires |
| Facial Recognition | Police surveillance | False arrest of innocent people | Model 35% less accurate on dark skin |
| GPT in Legal Filings | ChatGPT court citations | Fabricated case citations submitted to court | LLM hallucination not disclosed to judge |
CAUTION: Key insight: AI systems don't create new biases — they amplify and automate existing biases at scale. A biased algorithm makes biased decisions for millions of cases per day.