The Alignment Problem
   HOME
*





The Alignment Problem
''The Alignment Problem: Machine Learning and Human Values'' is a 2020 non-fiction book by the American writer Brian Christian. It is based on numerous interviews with experts trying to build artificial intelligence systems, particularly machine learning systems, that are aligned with human values. Summary The book is divided into three sections: Prophecy, Agency, and Normativity. Each section covers researchers and engineers working on different challenges in the alignment of artificial intelligence with human values. Prophecy In the first section, Christian interweaves discussions of the history of artificial intelligence research, particularly the machine learning approach of artificial neural networks such as the Perceptron and AlexNet, with examples of how AI systems can have unintended behavior. He tells the story of Julia Angwin, a journalist whose ProPublica investigation of the COMPAS algorithm, a tool for predicting recidivism among criminal defendants, led to wides ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Brian Christian
Brian Christian (born 1984 in Wilmington, Delaware) is an American non-fiction author, poet, programmer and researcher, best known for a bestselling series of books about the human implications of computer science, including ''The Most Human Human'' (2011), ''Algorithms to Live By'' (2016), and '' The Alignment Problem'' (2020). Christian competed as a "confederate" in the 2009 Loebner Prize competition, attempting to seem "more human" than the humans taking the test, and succeeded. The book he wrote about the experience, ''The Most Human Human,'' became a ''Wall Street Journal'' best-seller, a ''New York Times'' editors' choice, and a ''New Yorker'' favorite book of the year. He was interviewed by Jon Stewart on ''The Daily Show'' on March 8, 2011. In 2010, Christian collaborated with film director Michael Langan on a short film adaptation of Christian's poem "Heliotropes," which was published in the final issue of ''Wholphin'' magazine. In 2016, Christian collaborated with co ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


picture info

AlphaGo
AlphaGo is a computer program that plays the board game Go (game), Go. It was developed by DeepMind Technologies a subsidiary of Google (now Alphabet Inc.). Subsequent versions of AlphaGo became increasingly powerful, including a version that competed under the name AlphaGo Master, Master. After retiring from competitive play, AlphaGo Master was succeeded by an even more powerful version known as AlphaGo Zero, which was completely Self-play (reinforcement learning technique), self-taught without learning from human games. AlphaGo Zero was then generalized into a program known as AlphaZero, which played additional games, including chess and shogi. AlphaZero has in turn been succeeded by a program known as MuZero which learns without being taught the rules. AlphaGo and its successors use a Monte Carlo tree search algorithm to find its moves based on knowledge previously acquired by machine learning, specifically by an artificial neural network (a deep learning method) by extensi ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


picture info

Nature (journal)
''Nature'' is a British weekly scientific journal founded and based in London, England. As a multidisciplinary publication, ''Nature'' features peer-reviewed research from a variety of academic disciplines, mainly in science and technology. It has core editorial offices across the United States, continental Europe, and Asia under the international scientific publishing company Springer Nature. ''Nature'' was one of the world's most cited scientific journals by the Science Edition of the 2019 ''Journal Citation Reports'' (with an ascribed impact factor of 42.778), making it one of the world's most-read and most prestigious academic journals. , it claimed an online readership of about three million unique readers per month. Founded in autumn 1869, ''Nature'' was first circulated by Norman Lockyer and Alexander Macmillan as a public forum for scientific innovations. The mid-20th century facilitated an editorial expansion for the journal; ''Nature'' redoubled its efforts in exp ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Kirkus Reviews
''Kirkus Reviews'' (or ''Kirkus Media'') is an American book review magazine founded in 1933 by Virginia Kirkus (1893–1980). The magazine is headquartered in New York City. ''Kirkus Reviews'' confers the annual Kirkus Prize to authors of fiction, nonfiction, and young readers' literature. ''Kirkus Reviews'', published on the first and 15th of each month; previews books before their publication. ''Kirkus'' reviews over 10,000 titles per year. History Virginia Kirkus was hired by Harper & Brothers to establish a children's book department in 1926. The department was eliminated as an economic measure in 1932 (for about a year), so Kirkus left and soon established her own book review service. Initially, she arranged to get galley proofs of "20 or so" books in advance of their publication; almost 80 years later, the service was receiving hundreds of books weekly and reviewing about 100. Initially titled ''Bulletin'' by Kirkus' Bookshop Service from 1933 to 1954, the title was ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Publishers Weekly
''Publishers Weekly'' (''PW'') is an American weekly trade news magazine targeted at publishers, librarians, booksellers, and literary agents. Published continuously since 1872, it has carried the tagline, "The International News Magazine of Book Publishing and Bookselling". With 51 issues a year, the emphasis today is on book reviews. The magazine was founded by bibliographer Bibliography (from and ), as a discipline, is traditionally the academic study of books as physical, cultural objects; in this sense, it is also known as bibliology (from ). English author and bibliographer John Carter describes ''bibliography ... Frederick Leypoldt in the late 1860s, and had various titles until Leypoldt settled on the name ''The Publishers' Weekly'' (with an apostrophe) in 1872. The publication was a compilation of information about newly published books, collected from publishers and from other sources by Leypoldt, for an audience of booksellers. By 1876, ''The Publishers' Weekly ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


picture info

The Wall Street Journal
''The Wall Street Journal'' is an American business-focused, international daily newspaper based in New York City, with international editions also available in Chinese and Japanese. The ''Journal'', along with its Asian editions, is published six days a week by Dow Jones & Company, a division of News Corp. The newspaper is published in the broadsheet format and online. The ''Journal'' has been printed continuously since its inception on July 8, 1889, by Charles Dow, Edward Jones, and Charles Bergstresser. The ''Journal'' is regarded as a newspaper of record, particularly in terms of business and financial news. The newspaper has won 38 Pulitzer Prizes, the most recent in 2019. ''The Wall Street Journal'' is one of the largest newspapers in the United States by circulation, with a circulation of about 2.834million copies (including nearly 1,829,000 digital sales) compared with ''USA Today''s 1.7million. The ''Journal'' publishes the luxury news and lifestyle magazine ' ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  




William MacAskill
William David MacAskill (; born 24 March 1987) is a Scottish philosopher and author, as well as one of the originators of the effective altruism movement. He is an Associate Professor in Philosophy and Research Fellow at the Global Priorities Institute at the University of Oxford, and Director of the Forethought Foundation for Global Priorities Research. MacAskill is also the co-founder of Giving What We Can, the Centre for Effective Altruism and 80,000 Hours. He is the author of the 2015 book ''Doing Good Better'', the 2022 book ''What We Owe the Future'', and co-author of the 2020 book ''Moral Uncertainty.'' Early life and education MacAskill was born William Crouch in 1987, and grew up in Glasgow. MacAskill was educated at Hutchesons' Grammar School in Glasgow. At the age of 15, after learning about how many people were dying as a result of AIDS, he made the decision to work towards becoming wealthy and giving away half of his money. At the age of 18, MacAskill read Peter ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Toby Ord
Toby David Godfrey Ord (born July 1979) is an Australian philosopher. He founded Giving What We Can in 2009, an international society whose members pledge to donate at least 10% of their income to effective charities, and is a key figure in the effective altruism movement, which promotes using reason and evidence to help the lives of others as much as possible. He is a senior research fellow at the University of Oxford's Future of Humanity Institute, where his work is focused on existential risk. His book on the subject '' The Precipice: Existential Risk and the Future of Humanity'' was published in March 2020. Early life and education Ord was born in Melbourne, Australia, in 1979. He later attended the University of Melbourne, where he initially studied computer science. On completing his first degree, he switched to studying philosophy to pursue his interest in ethics, later stating: "At this stage I knew that I wanted to make a large positive difference in the world and it ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


picture info

Existential Risk
A global catastrophic risk or a doomsday scenario is a hypothetical future event that could damage human well-being on a global scale, even endangering or destroying modern civilization. An event that could cause human extinction or permanently and drastically curtail humanity's potential is known as an "existential risk." Over the last two decades, a number of academic and non-profit organizations have been established to research global catastrophic and existential risks, formulate potential mitigation measures and either advocate for or implement these measures. Definition and classification Defining global catastrophic risks The term global catastrophic risk "lacks a sharp definition", and generally refers (loosely) to a risk that could inflict "serious damage to human well-being on a global scale". Humanity has suffered large catastrophes before. Some of these have caused serious damage but were only local in scope—e.g. the Black Death may have resulted in the de ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Effective Altruism
Effective altruism is a philosophical and social movement that advocates "using evidence and reason to figure out how to benefit others as much as possible, and taking action on that basis". People who pursue the goals of effective altruism, called , often choose careers based on the amount of good that the career achieves while donating to charities based on maximising impact. The movement developed during the 2000s, and the name was coined in 2011. Prominent philosophers influential to the movement include Peter Singer, Toby Ord, and William MacAskill. Several books and many articles about the movement have since been published, and the Effective Altruism Global conference has been held since 2013. As of 2022, several billion dollars have been committed to effective altruist causes. Popular cause priorities within effective altruism include global health and development, social inequality, animal welfare, and risks to the survival of humanity over the long-term future. Eff ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


Normative
Normative generally means relating to an evaluative standard. Normativity is the phenomenon in human societies of designating some actions or outcomes as good, desirable, or permissible, and others as bad, undesirable, or impermissible. A norm in this normative sense means a standard for evaluating or making judgments about behavior or outcomes. Normative is sometimes also used, somewhat confusingly, to mean relating to a descriptive standard: doing what is normally done or what most others are expected to do in practice. In this sense a norm is not evaluative, a basis for judging behavior or outcomes; it is simply a fact or observation about behavior or outcomes, without judgment. Many researchers in science, law, and philosophy try to restrict the use of the term normative to the evaluative sense and refer to the description of behavior and outcomes as positive, descriptive, predictive, or empirical. ''Normative'' has specialised meanings in different academic disciplines such a ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]  


picture info

Inverse Reinforcement Learning
Reinforcement learning (RL) is an area of machine learning concerned with how intelligent agents ought to take actions in an environment in order to maximize the notion of cumulative reward. Reinforcement learning is one of three basic machine learning paradigms, alongside supervised learning and unsupervised learning. Reinforcement learning differs from supervised learning in not needing labelled input/output pairs to be presented, and in not needing sub-optimal actions to be explicitly corrected. Instead the focus is on finding a balance between exploration (of uncharted territory) and exploitation (of current knowledge). The environment is typically stated in the form of a Markov decision process (MDP), because many reinforcement learning algorithms for this context use dynamic programming techniques. The main difference between the classical dynamic programming methods and reinforcement learning algorithms is that the latter do not assume knowledge of an exact mathematica ...
[...More Info...]      
[...Related Items...]     OR:     [Wikipedia]   [Google]   [Baidu]