Trustworthy AI: A Business Guide for Navigating Trust and Ethics in AI

Cover
Title Page
Copyright Page
Contents
Foreword
Preface
Acknowledgments
Introduction
	Navigating the Innovation Curve
	The Dimensions of Trustworthy AI
Chapter 1 A Primer on Modern AI
	The Road to Machine Intelligence
	Basic Terminology in AI
	Types of AI Models and Use Cases
	New Challenges for the Modern AI Era
Chapter 2 Fair and Impartial
	A Longstanding Ethical Question
	The Nature of Bias in AI
		Selection Bias
		Confirmation Bias
		Explicit and Implicit Bias
		Institutional Bias
	Tradeoffs in Fairness
	Leading Practices in Promoting Fairness
		Build a Diverse Team
		Balance the Datasets
		Probe the Data
		Engage AI Stakeholders
		Develop Pathways for Feedback and Evaluating Performance
	Toward a Fairer Future in AI
Chapter 3 Robust and Reliable
	Robust vs Brittle AI
	Developing Reliable AI
	The Challenge of Generalizable Deep Learning
	Factors Influencing AI Reliability
		Lessons in Data Reliability
		Meeting the Long Tail in Reliability Engineering
	Robustness and Bad Actors
	Consequences Worth Contemplating
	Leading Practices in Building Robust and Reliable AI
		Benchmarks for Reliability
		Perform Data Audits
		Monitor Reliability Over Time
		Uncertainty Estimates
		Managing Drift
		Continuous Learning
		Ongoing Testing
		Explore Alternative Approaches
	Driving Toward Robust and Reliable Tools
Chapter 4 Transparent
	Defining the Nature of Transparency in AI
	The Limits of Transparency
	Weighing the Impact on the Stakeholders
	Taking Steps into Transparency
		People
		Processes
		Technology
	Trust from Transparency
Chapter 5 Explainable
	The Components of Understanding AI Function
	The Value in Explainable AI
		Driving Innovation and Application
		Improving Model Performance
		Encouraging Use Through Trust
		Satisfying Regulatory Inquiries
	Factors in Explainability
	Technical Approaches to Fostering Explainability
		Prioritizing Explainability
		Intrinsic Explainability
		Post Hoc Explainability
	Leading Practices in Process
		Engage All Stakeholders
		Tailor Explanations and Reporting to the Stakeholder
		Ongoing Explainability Testing
	The Explainable Imperative
Chapter 6 Secure
	What Does AI Compromise Look Like?
	How Unsecure AI Can Be Exploited
		Data Poisoning
		Transfer Learning Attack
		Reverse Engineering the Code
		Exploiting System Errors
	The Consequences from Compromised AI
		Data Exposure
		Loss of Intellectual Property
		Bypassing Filters
		Liability and Regulatory Fines
		User Trust in AI
	Leading Practices for Shoring-Up AI Security
		People
		Processes
		Technology
	Securing the Future with AI
Chapter 7 Safe
	Understanding Safety and Harm in AI
		Physical Harm
		Psychological Harm
		Economic Harm
		Environmental Harm
		Legal Harm
		Optimizing for Human Values
	Aligning Human Values and AI Objectives
	Technical Safety Leading Practices
		Set Safety Metrics for Assessment
		Perform Monitoring and Assessments
		Bring Everyone to the Table
		Understand the Values of the Business and the End User
		Build and Deploy for the Future
	Seeking a Safer Future with AI
Chapter 8 Privacy
	Consent, Control, Access, and Privacy
	The Friction Between AI Power and Privacy
		Unobvious Data Collection
		Predicting and Inferring Details
		Biometrics and Behavior
		Traditional Technology with New Capabilities
	Beyond Anonymization or Pseudonymization
	Privacy Laws and Regulations
		General Data Protection Regulation (GDPR)
		California Consumer Privacy Act (CCPA) and California Privacy Rights Act (CPRA)
		Other Notable Privacy Laws
	Leading Practices in Data and AI Privacy
		Supplementing the Datasets
		Obtain Informed Consent
		Define Privacy Objectives
		Engage Stakeholders
	The Nexus of AI Trust and Privacy
Chapter 9 Accountable
	Accountable for What and to Whom?
	Balancing Innovation and Accountability
	Laws, Lawsuits, and Liability
	Leading Practices in Accountable AI
		People
		Processes
		Technology
	Accounting for Trust in AI
Chapter 10 Responsible
	Corporate Responsibility in the AI Era
	Motivating Responsible AI Use
	Balancing Good, Better, and Best
	Leading Practices in the Responsible Use of AI
		Lead with Principled Strategy
		Charge an AI Advisory Board with Oversight
		Intentionally Build Diversity
		Define Processes for AI Assessments
	Trust Emerging from Responsibility
Chapter 11 Trustworthy AI in Practice
	Step 1 – Identify the Relevant Dimensions of Trust
	Step 2 – Cultivating Trust Through People, Processes, and Technologies
		People
		Processes
		Technology
	Guidelines for Action on Trustworthy AI
		Principles
		Contributing Factors
		Human Capital Development
	Taking the Next Steps
Chapter 12 Looking Forward
Notes
	Introduction
	Chapter 1
	Chapter 2
	Chapter 3
	Chapter 4
	Chapter 5
	Chapter 6
	Chapter 7
	Chapter 8
	Chapter 9
	Chapter 10
	Chapter 11
	Chapter 12
Index
EULA