Today, machine learning systems are used to make decisions for us, from selecting resumes to providing medical care. However, these systems often exhibit biases and create ethical issues. When the systems ultimately do not perform as we expect, ethical and potentially existential risks arise. Researchers refer to this issue as the alignment problem.

In Brian Christian's compelling work, we encounter the first mentions of the alignment problem and learn about the ambitious plans to solve it. Through a unique combination of historical overview and on-the-ground journalism, Christian tracks the explosive growth of the machine learning field, highlighting an industry that is finding its footing between exciting and, at times, frightening progress.

Whether we will succeed or fail in solving the alignment problem will become one of humanity's defining stories.

REVIEWS

"The divergence between intent and outcome―between what mathematician Norbert Wiener described as 'the purpose embedded in the machine' and 'the purpose we truly desire'―defines the essence of the 'alignment problem.' Brian Christian, an acclaimed technology writer, offers a detailed and captivating exploration of this hot issue while providing an overview of the state of machine learning and the challenges it faces."

― David A. Shaywitz, Wall Street Journal

"A fascinating, thought-provoking, and insightful tour of all the ways AI fails and all the efforts to correct it. A must-read for anyone wanting to understand where our world is headed."

― Stuart Russell, professor of computer science, University of California, Berkeley, and author of Human Compatible

"A new domain has emerged that responds to and examines the massive technological changes represented by our modern, virtual, algorithmically defined world. In 'The Alignment Problem,' Brian Christian masterfully chronicles the 'AI fairness' community, introducing some of its key figures, its historical roots in science, philosophy, and activism, and, most importantly, many of its philosophical dilemmas and limitations."

― Cathy O’Neil, author of Weapons of Math Destruction

"This is the book on artificial intelligence that we need right now. Brian Christian leads us on a technically astute (yet accessible) journey through the most significant questions facing AI and humanity. It is a challenging and vital read for anyone concerned about our future."

― Mike Krieger, co-founder of Instagram

"A richly researched and compelling book that explores the path humanity has taken to create a successor for itself―a path filled with surprises, unexpected hurdles, clever solutions, and, increasingly, tough questions about the soul of our species."

― Jaan Tallinn, co-founder of Skype and the Future of Life Institute

"'The Alignment Problem' should be required reading for anyone influencing policy where algorithms are involved―that is, everywhere. But, unlike many required readings, this book is a joy, a playful journey through personalities and relevant moments in scientific history that place today's choices in the right context."

― Jennifer Pahlka, founder of Code for America and former US Deputy CTO

"A deeply enjoyable and thoroughly researched account of how computer scientists and philosophers are defining the greatest question of our time: how will we create intelligent machines that enhance our lives instead of complicating or even destroying them? There is no better book than 'The Alignment Problem' to explain the issues of safe governance of AI."

― James Barrat, best-selling author of Our Final Invention

"Brian Christian is an outstanding writer and has created a captivating book. AI appears to be destined, for better or worse, to take on an increasingly central role in our lives. We should be grateful for this balanced and unexaggerated perspective on its scope and limitations."

― Martin Rees, emeritus professor of cosmology and astrophysics, University of Cambridge

"An intriguing exploration of AI, which is advancing faster than―well, than we are."

― Kirkus Reviews