Lecture 26: Ethics and Trust in AI

Content locked

The catalog is public. Reading this content requires approved access.

Introduction

The AI Knowledge-Ethics Matrix

Dual Use of AI

Bias and Fairness

The Problem of Algorithmic Bias

Fairness Through Randomization

Real-World Examples of Bias

Causes of Bias in ML

Mitigating Bias

Group Fairness Definitions

Fake Content and Misinformation

Deepfake Videos

AI-Generated Images

AI-Generated Text

Detecting Fake Content

Privacy

The Netflix Prize Case Study

Right to be Forgotten

Differential Privacy

Adversarial Robustness

Adversarial Examples

Physical Attacks

Adversarial Examples in NLP

Test-Time Attacks

Defense: Adversarial Training

Practice Problems

Key Takeaways