The Alignment Problem: Machine Learning and Human Values Paperback - 2021
by Christian, Brian
- New
- Paperback
Standard delivery: 4 to 14 days
Details
- Title The Alignment Problem: Machine Learning and Human Values
- Author Christian, Brian
- Binding Paperback
- Condition New
- Pages 496
- Volumes 1
- Language ENG
- Publisher W. W. Norton & Company
- Publication date 2021-10-05
- Bookseller's Inventory # SKU0615790
- ISBN 9780393868333 / 0393868338
- Weight 0.85 lbs (0.39 kg)
- Dimensions 8.19 x 5.43 x 1.26 in (20.80 x 13.79 x 3.20 cm)
- Size 5x1x8
- Category Science
- Quantity available 2
About SGS Trading Inc New Jersey, United States
Textbook and Reference Books Discounted
Reader reviews for The Alignment Problem: Machine Learning and Human Values
Review summary
Readers broadly praised The Alignment Problem as a deeply researched, lucid survey of modern machine learning and its ethical pitfalls, weaving clear explanations with vivid anecdotes and insights from psychology and neuroscience. Many lauded its treatment of bias, brittleness, and reinforcement learning, and appreciated its balanced, cautiously optimistic stance on AI safety. Criticisms centered on length and structure—several found it meandering or more a general ML compendium than a tight study of alignment—and some noted parts feel dated or light on practical solutions. Despite caveats, it’s widely recommended as an accessible yet substantive primer for newcomers and a stimulating synthesis for practitioners.
Readers say this book is:
well-researchedinformation-denseaccessiblethought-provokinginterdisciplinaryengagingcautiously optimisticunevenoverlongoccasionally datedWrite a review for this book
Important Terms and Guidelines
- Please focus on the book’s content and context. Also, add any personal comments as to how you enjoyed the book. Substantiate your likes and dislikes. You may make comparisons to other books.
- Reviews must be at least 140 characters in length.
- Please do not reveal critical plot elements.
- This is not a help line. Contact customer support if you need help.
Your review must not include:
- Obscenities, discriminatory language, or other insulting language not suitable for public domain
- Advertisements, “spam” content, or references to other products, offers or websites.
- Email addresses, URLs, phone numbers, physical addresses or other contact information.
- Overly critical comments about other reviews or reviewers
- Time-sensitive material (i.e. promotional tours, seminars, lectures, etc.)
- Availability, price, or alternative ordering/shipping information