AI Accountability Essential Training

Learn why it's absolutely crucial for AI-related data science work to be transparent, explainable, accountable, and ethical in its design and execution.

Introduction

• What is AI accountability?

1. The Context for AI

• The promise of AI
• General and narrow AI

2. Technical Challenges of AI

• The challenge of classification errors
• The causes of classification errors
• Bias in AI
• Supervised and unsupervised learning
• Biased labeling of data
• Construct validity
• The absence of meaning
• Vulnerability to attacks

3. Social Challenges of AI

• Dimensions of justice
• Moral and relational reasoning
• Issues of authenticity

4. Legal Challenges of AI

• Privacy laws
• Spurious discrimination
• The right to explanation
• Discrimination in data
• Discrimination in implementation

5. Safety Challenges of AI

• AI in life and death situations
• AI in the military
• The challenges of military AI

6. Confronting the Challenges of AI

• Strategies for developers
• Strategies for executives
• Strategies for public relations
• Strategies for regulators
• Strategies for consumers

Conclusion

• Next steps