Acknowledgements
Contents
Contributors
Chapter 1: Introduction - The Importance of an Ethics-First Approach to the Development of AI
References
Chapter 2: A Unified Framework of Five Principles for AI in Society
2.1 Introduction
2.2 Artificial Intelligence: A Research Area in Search of a Definition
2.3 A Unified Framework of Five Principles for Ethical AI
2.3.1 Beneficence: Promoting Well-Being, Preserving Dignity, and Sustaining the Planet
2.3.2 Non-maleficence: Privacy, Security and Capability Cautionรยด
2.3.3 Autonomy: The Power to Decide (to Decide)
2.3.4 Justice: Promoting Prosperity, Preserving Solidarity, Avoiding Unfairness
2.3.5 Explicability: Enabling the Other Principles Through Intelligibility and Accountability
2.3.6 A Synoptic View
2.4 AI Ethics: Whence and for Whom?
2.5 Conclusion: From Principles to Practices
References
Chapter 3: An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations
3.1 Introduction
3.2 The Opportunities and Risks of AI for Society
3.2.1 Who We Can Become: Enabling Human Self-Realisation, Without Devaluing Human Abilities
3.2.2 What We Can Do: Enhancing Human Agency, Without Removing Human Responsibility
3.2.3 What We Can Achieve: Increasing Societal Capabilities, Without Reducing Human Control
3.2.4 How We Can Interact: Cultivating Societal Cohesion, Without Eroding Human Self-Determination
3.3 The Dual Advantage of an Ethical Approach to AI
3.4 A Unified Framework of Principles for AI in Society
3.4.1 Beneficence: Promoting Well-Being, Preserving Dignity, and Sustaining the Planet
3.4.2 Non-maleficence: Privacy, Security and ``Capability Cautionรยดรยด
3.4.3 Autonomy: The Power to Decide (Whether to Decide)
3.4.4 Justice: Promoting Prosperity and Preserving Solidarity
3.4.5 Explicability: Enabling the Other Principles Through Intelligibility and Accountability
3.5 Recommendations for a Good AI Society
3.5.1 Preamble
3.5.2 Action Points
3.5.2.1 Assessment
3.5.2.2 Development
3.5.2.3 Incentivisation
3.5.2.4 Support
3.6 Conclusion
References
Chapter 4: Establishing the Rules for Building Trustworthy AI
4.1 Careful Planning Rather Than Beta Testing
4.2 Ethics First to Inform Legislation
4.3 Further Steps for a Global Stage
References
Chapter 5: The Chinese Approach to Artificial Intelligence: An Analysis of Policy, Ethics, and Regulation
5.1 Introduction
5.2 AI Governance in China
5.2.1 The New Generation Artificial Intelligence Development Plan (AIDP)
5.2.2 Implementing the AIDP
5.3 Chinaรยดs AI Strategic Focus
5.3.1 International Competition
5.3.2 Economic Development
5.3.3 Social Governance
5.3.4 Moral Governance
5.4 The Debate on Digital Ethics and AI in China
5.4.1 Privacy
5.4.2 Medical Ethics
5.5 Conclusion
References
Chapter 6: Translating Principles into Practices of Digital Ethics: Five Risks of Being Unethical
6.1 Introduction
6.2 Ethics Shopping
6.3 Ethics Bluewashing
6.4 Ethics Lobbying
6.5 Ethics Dumping
6.6 Ethics Shirking
6.7 Conclusion
References
Chapter 7: How AI Can Be a Force for Good - An Ethical Framework to Harness the Potential of AI While Keeping Humans in Control
7.1 Delegation and Responsibility
7.2 Invisibility and Influence
7.3 Translational Ethics
References and Notes
Chapter 8: The Ethics of Algorithms: Key Problems and Solutions
8.1 Introduction
8.2 Map of the Ethics of Algorithms
8.3 Inconclusive Evidence Leading to Unjustified Actions
8.4 Inscrutable Evidence Leading to Opacity
8.5 Misguided Evidence Leading to Unwanted Bias
8.6 Unfair Outcomes Leading to Discrimination
8.7 Transformative Effects Leading to Challenges for Autonomy and Informational Privacy
8.8 Traceability Leading to Moral Responsibility
8.9 Conclusion
References
Chapter 9: How to Design AI for Social Good: Seven Essential Factors
9.1 Introduction
9.2 Methodology
9.3 Seven Essential Factors for Successful AI4SG
9.3.1 Falsifiability and Incremental Deployment
9.3.2 Safeguards Against the Manipulation of Predictors
9.3.3 Receiver-Contextualised Intervention
9.3.4 Receiver-Contextualised Explanation and Transparent Purposes
9.3.5 Privacy Protection and Data Subject Consent
9.3.6 Situational Fairness
9.3.7 Human-Friendly Semanticisation
9.4 Conclusion: Balancing Factors for AI for Social Good
Appendix: Representative AI4SG Examples
References
Chapter 10: From What to How: An Initial Review of Publicly Available AI Ethics Tools, Methods and Research to Translate Princ...
10.1 Introduction
10.2 Moving from Principles to Practices
10.3 Methodology
10.4 Framing the Results
10.5 Discussion of Initial Results
10.5.1 Explicability as the All-Encompassing Principle
10.5.2 An Individual Focus
10.5.3 A Lack of Usability
10.6 A Way Forward
10.7 Limitations
10.8 Conclusion
References
Chapter 11: The Explanation Game: A Formal Framework for Interpretable Machine Learning
11.1 Introduction
11.2 Why Explain Algorithms?
11.2.1 Justice as (Algorithmic) Fairness
11.2.2 The Context of (Algorithmic) Justification
11.2.3 The Context of (Algorithmic) Discovery
11.3 Formal Background
11.3.1 Supervised Learning
11.3.2 Causal Interventionism
11.3.3 Decision Theory
11.4 Scope
11.4.1 Complete
11.4.2 Precise
11.4.3 Forthcoming
11.5 The Explanation Game
11.5.1 Three Desiderata
11.5.1.1 Accuracy
11.5.1.2 Simplicity
11.5.1.3 Relevance
11.5.2 Rules of the Game
11.5.2.1 Inputs
11.5.2.2 Mapping the Space
11.5.2.3 Building Models, Scoring Explanations
11.5.3 Consistency and Convergence
11.6 Discussion
11.7 Objections
11.7.1 Too Highly Idealised
11.7.2 Infinite Regress
11.7.3 Pragmatism + Pluralism = Relativist Anarchy?
11.7.4 No Trade-Off
11.7.5 Double Standards
11.8 Conclusion
References
Chapter 12: Artificial Agents and Their Moral Nature
12.1 Introduction: Standard vs. Non-standard Theories of Agents and Patients
12.2 What Is an Agent?
12.2.1 On the Very Idea of Levels of Abstraction
12.2.2 Definitions
12.2.3 Relativism
12.2.4 State and State-Transitions
12.2.5 An Effective Characterisation of Agents
12.2.6 Examples
12.2.6.1 The Defining Properties
12.2.6.2 Noughts and Crosses
12.2.6.3 Webbot
12.2.6.4 Futuristic Thermostat
12.2.6.5 SmartPaint
12.2.6.6 Organisations
12.3 Morality
12.3.1 Morality of Agents
12.3.2 A-Responsible Morality
12.3.2.1 The Teleological Objection
12.3.2.2 The Intentional Objection
12.3.2.3 The Freedom Objection
12.3.2.4 The Responsibility Objection
12.3.3 Morality Threshold
12.3.3.1 Examples
12.4 Information Ethics
12.4.1 Codes of Ethics
12.4.2 Censorship
12.5 Conclusion
References
Chapter 13: Artificial Intelligence Crime: An Interdisciplinary Analysis of Foreseeable Threats and Solutions
13.1 Introduction
13.2 Methodology
13.3 Threats
13.3.1 Commerce, Financial Markets, and Insolvency
13.3.2 Harmful or Dangerous Drugs
13.3.3 Offences Against the Person
13.3.4 Sexual Offences
13.3.5 Theft and Fraud, and Forgery and Personation
13.4 Possible Solutions for Artificial Intelligence-Supported Crime
13.4.1 Tackling Emergence
13.4.2 Addressing Liability
13.4.3 Monitoring
13.4.4 Psychology
13.5 Conclusions
References
Chapter 14: Regulate Artificial Intelligence to Avert Cyber Arms Race
14.1 No Rules
14.2 Three Steps
References
Chapter 15: Trusting Artificial Intelligence in Cybersecurity Is a Double-Edged Sword
15.1 Vulnerabilities of AI
15.2 Standards and Certification Procedures
15.3 Making AI in Cybersecurity Reliable
References
Chapter 16: Prayer-Bots and Religious Worship on Twitter: A Call for a Wider Research Agenda
16.1 Introduction
16.2 Islamic Prayer Apps
16.3 Religious Context
16.4 Broader Implications
References
Chapter 17: Artificial Intelligence, Deepfakes and a Future of Ectypes
References
Chapter 18: The Ethics of AI in Health Care: A Mapping Review
18.1 Introduction
18.2 Methodology
18.3 Findings
18.3.1 Epistemic Concerns: Inconclusive, Inscrutable, and Misguided Evidence
18.3.2 Normative Concerns: Unfair Outcomes and Transformative Effects
18.3.3 Overarching Concerns: Traceability
18.4 The Need for an Ethically-Mindful and Proportionate Approach
18.5 Conclusion
Appendix - Methodology
References
Chapter 19: Autonomous Vehicles: From Whether and When to Where and How
19.1 Introduction
19.2 Travelling, Journeys and Trips
19.3 NotWhenรยด or Even Whereรยด ButHowรยด Is the Question
19.4 Conclusion
References
Chapter 20: Innovating with Confidence: Embedding AI Governance and Fairness in a Financial Services Risk Management Framework
20.1 Introduction
20.2 Fairness in the Financial Services Industry
20.3 Managing Risks of AI Through Its Lifecycle
20.4 Design
20.4.1 Definition of Scope
20.4.2 Risk Identification and Assessment
20.4.3 Risk Management Plan and Control Design
20.4.4 Defined Roles and Responsibilities
20.5 Build
20.5.1 Development and Testing Process
20.5.2 Governance and Oversight
20.5.3 Documented Change Management, Testing, and Approval
20.5.4 Transparency and Explainability
20.6 Productionize
20.6.1 Ensuring Solution Is Safe to Scale
20.6.2 Review the Feedback Mechanism
20.6.3 ``Kill Switchรยดรยด and Business Continuity
20.7 Monitor
20.7.1 Automated Monitoring and Testing
20.7.2 Vulnerable Customers
20.7.3 Periodic Re-Validation
20.7.4 Internal Audit Planning
20.8 Conclusion
20.9 Funding
Bibliography
Chapter 21: Robots, Jobs, Taxes, and Responsibilities
References
Chapter 22: What the Near Future of Artificial Intelligence Could Be
22.1 Introduction
22.2 AI: A Working Definition
22.3 Alรยดs Future: From Historical Data to Hybrid and Synthetic Data, and the Need for Ludification
22.4 Alรยดs Future: From Difficult Problems to Complex Problems, and the Need for Enveloping
22.5 Conclusion: A Future of Design
References