Join the AISES course
Take action
Course overview
Curriculum
Download textbook
1. Overview of Catastrophic AI Risks
0.1
Preface
1.1
Overview of Catastrophic AI Risks
1.2
Malicious Use
1.3
AI Race
1.4
Organizational Risks
1.5
Rogue AIs
1.6
Discussion of Connections Between Risks
2. AI Fundamentals
2.1
AI Fundamentals
2.2
Artificial Intelligence & Machine Learning
2.3
Deep Learning
2.4
Scaling Laws
2.5
Speed of AI Development
2.6
AI Fundamentals Conclusion
3. Single Agent Safety
3.1
Single Agent Safety
3.2
Monitoring
3.3
Robustness
3.4
Alignment
3.5
Systemic Safety
3.6
Safety and General Capabilities
3.7
Conclusion
4. Safety Engineering
4.1
Safety Engineering
4.2
Risk Decomposition
4.3
Nines of Reliability
4.4
Safe Design Principles
4.5
Component Failure Accident Models and Methods
4.6
Systemic Factors
4.7
Tail Events and Black Swans
4.8
Conclusion
5. Complex Systems
5.1
Complex Systems
5.2
Introduction to Complex Systems
5.3
Complex Systems for AI Safety
5.4
Conclusion
6. Beneficial AI and Machine Ethics
6.1
Beneficial AI and Machine Ethics
6.2
Law
6.3
Fairness
6.4
The Economic Engine
6.5
Wellbeing
6.6
Preferences
6.7
Happiness
6.8
Social Welfare Functions
6.9
Moral Uncertainty
7. Collective Action Problems
7.1
Collective Action Problems
7.2
Game Theory
7.3
Cooperation
7.4
Conflict
7.5
Evolutionary Pressures
7.6
Conclusion
8. Governance
8.1
Governance
8.2
Growth
8.3
Distribution
8.4
Corporate Governance
8.5
National Governance
8.6
International Governance
8.7
Compute Governance
8.8
Conclusion
9. Appendices
9.1
App. A: Normative Ethics
10.1
App. B: Utility Functions
11.1
App. C: Reinforcement Learning
12.1
App. D: Long-Tailed and Thin-Tailed Distributions
13.1
App. E: Evolutionary Game Theory
14.1
App. F: Other Cooperation Mechanisms
15.1
App. G: Intrasystem Conflict Causes
16.1
Acknowledgements
All
Section
Appendix
2.2

Artificial Intelligence & Machine Learning

Artificial intelligence involves developing computer systems that can perform tasks that typically require human intelligence, from visual perception to decision-making. Machine learning is an approach to AI that involves developing models that can learn from data to perform tasks without being explicitly programmed.

No items found.

Review Questions

What is artificial intelligence (AI)?

Answer:

Artificial intelligence (AI) is an umbrella term that encapsulates the overarching idea of creating machines that perform tasks typically associated with human intelligence. It lacks a universal definition and is used to refer to many different—but related—ideas.

‍

View Answer
Hide Answer

Several conceptual categories have been developed and are often used to describe the capabilities of AI systems including, but not limited to, narrow AI, artificial general intelligence (AGI), human-level AI (HLAI), transformative AI (TAI), and superintelligence (ASI). For each of the following descriptions, name the type of AI to which it corresponds and provide an example.

  1. AI systems with impacts that are comparable to or more significant than the Industrial Revolution.
  2. AI systems that can learn and perform a wide range of tasks in many different areas.
  3. AI systems that significantly outclasses human performance in virtually all domains of interest.
  4. AI systems that are designed to perform specific tasks or solve particular problems within a specialized domain of expertise.
  5. AI systems that can perform approximately every task as well (and, in some definitions, at least as cheaply) as human workers.

‍

Answer:

Types of AI:

  1. Transformative AI (TAI)
  2. Artificial general intelligence (AGI)
  3. Superintelligence (ASI)
  4. Narrow AI
  5. Human-level artificial intelligence (HLAI)

‍

View Answer
Hide Answer

What is machine learning (ML)? What distinguishes ML from AI?

‍

Answer:

Machine learning (ML) is a subfield of AI that focuses on developing computer systems that can learn directly from data without following explicit pre-set instructions. ML is distinct from AI in that its systems use pattern recognition to extract information from and adapt to data.

‍

View Answer
Hide Answer
0.1
1.1
1.2
1.3
1.4
1.5
1.6
2.1
2.2
2.3
2.4
2.5
2.6
3.1
3.2
3.3
3.4
3.5
3.6
3.7
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
5.1
5.2
5.3
5.4
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
7.1
7.2
7.3
7.4
7.5
7.6
8.1
8.2
8.3
8.4
8.5
8.6
8.7
8.8
9.1
10.1
11.1
12.1
13.1
14.1
15.1
16.1
Get Involved
Express InterestVirtual courseFacilitateTake ActionSubmit Feedback
Textbook Resources
Start reading
Textbook audiobook
Table of contentsDownload textbookContent for curriculum
Terms of ServicePrivacy Policy

Citation:
Dan Hendrycks. Introduction to AI Safety, Ethics and Society. Taylor & Francis, (2024). ISBN: 9781032798028. URL: www.aisafetybook.com

Cookies Notice: This website uses cookies to identify pages that are being used most frequently. This helps us analyze data about web page traffic and improve our website. We only use this information for the purpose of statistical analysis and then the data is removed from the system. We do not and will never sell user data. Read more about our cookie policy on our privacy policy. Please contact us if you have any questions.

© 2024 Center for AI Safety
Built by Osborn Design Works