Human Compatible: Artificial Intelligence and the Problem of Control

Stuart Russell

Penguin, Oct 8, 2019 - Technology & Engineering - 352 pages

"The most important book on AI this year." --The Guardian

"Mr. Russell's exciting book goes deep, while sparkling with dry witticisms." --The Wall Street Journal

"The most important book I have read in quite some time" (Daniel Kahneman); "A must-read" (Max Tegmark); "The book we've all been waiting for" (Sam Harris)

A leading artificial intelligence researcher lays out a new approach to AI that will enable us to coexist successfully with increasingly intelligent machines

In the popular imagination, superhuman artificial intelligence is an approaching tidal wave that threatens not just jobs and human relationships, but civilization itself. Conflict between humans and machines is seen as inevitable and its outcome all too predictable.

In this groundbreaking book, distinguished AI researcher Stuart Russell argues that this scenario can be avoided, but only if we rethink AI from the ground up. Russell begins by exploring the idea of intelligence in humans and in machines. He describes the near-term benefits we can expect, from intelligent personal assistants to vastly accelerated scientific research, and outlines the AI breakthroughs that still have to happen before we reach superhuman AI. He also spells out the ways humans are already finding to misuse AI, from lethal autonomous weapons to viral sabotage.

If the predicted breakthroughs occur and superhuman AI emerges, we will have created entities far more powerful than ourselves. How can we ensure they never, ever, have power over us? Russell suggests that we can rebuild AI on a new foundation, according to which machines are designed to be inherently uncertain about the human preferences they are required to satisfy. Such machines would be humble, altruistic, and committed to pursue our objectives, not theirs. This new foundation would allow us to create machines that are provably deferential and provably beneficial.

What people are saying

LibraryThing Review

User Review  - fpagan - LibraryThing

Here's a book from which any reader can get an authentic glimpse of current AI research from a practicing expert in the field. The prospect that true artificial superintelligence may someday be ...

User Review -

I am confused by the book. If there is a belief that computers will become superintelligent, what makes the author believe simple coding for human preferences will make them manageable (or for that matter, complex coding)? This is a great book but I don't see why machines cannot achieve sentience like humans. And if machines achieve sentience, what will keep them from saying "hey, why am I wasting my time serving these idiot humans?" I suspect that superintelligence - a so-called singularity will be the end of biological intelligence as we know it. As much as I hate to admit it, superintelligence may be the way life evolves -- evolution of intelligent humans that gives way to "living" machines.
I suspect that free will is a farce. The delay between a decision and carrying it out gives way to a sense of free will. We already know that decisions are made in our minds before we are even cognizant of the decision (see Frank Wilczek's latest book). We are biological machines subject to the whims of the data from the outside world and the DNA/structure of our brains. We have something called sentience probably because we have a delay in our decision mechanisms. Why would we think that machines could not be made to be the same way -- or if the singularity is achieved, make themselves achieve the same thing.  

Contents

• chapter: 1
• chapter: 7
• INTELLIGENCE IN HUNAANS AND MACHINES: 13
• How MIGHT AI PROGRESS IN THE FUTURE2: 62
• NAISUSES OF AI: 103
• OVERLY INTELLIGENT Al: 132
• THE NOTSOGREAT AI DEBATE: 145
• A DIFFERENT APPROACH: 171
• PROVABLY BENEFICIAL Al: 184
• US: 211
• PROBLENA SOLVED?: 246
• SEARCHING FOR SOLUTIONS: 257
• KNOWLEDGE AND LOGIC: 267
• UNCERTAINTY AND PROBABILITY: 273
• LEARNING FRONT EXPERIENCE: 285
• Acknowledgments: 297
• Image Credits: 324

About the author

Stuart Russell is a professor of Computer Science and holder of the Smith-Zadeh Chair in Engineering at the University of California, Berkeley. He has served as the Vice-Chair of the World Economic Forum's Council on AI and Robotics and as an advisor to the United Nations on arms control. He is a Fellow of the American Association for Artificial Intelligence, the Association for Computing Machinery, and the American Association for the Advancement of Science. He is the author (with Peter Norvig) of the definitive and universally acclaimed textbook on AI, Artificial Intelligence: A Modern Approach.

Bibliographic information

• Title: Human Compatible: Artificial Intelligence and the Problem of Control
• Author: Stuart Russell
• Publisher: Penguin, 2019
• ISBN: 0525558624, 9780525558620
• Length: 352 pages
• Subjects: Business & Economics / Industries / Computers & Information Technology; Science / Biotechnology; Technology & Engineering / Robotics