What do you think?
Rate this book
Ai: The Tumultuous History Of The Search For Artificial Intelligence
In the summer of 1956, ten young scientists, some barely out of their doctoral studies, sat down to consider the astounding proposition that "every aspect of learning or any other feature of intelligence can, in principle, be so precisely described that a machine can be made to simulate it." Armed with their own enthusiasm, the excitement of the idea itself, and an infusion of government money, they predicted that the whole range of human intelligence would be programmable within their own lifetimes. Nearly half a century later, the field has grown exponentially - with mixed results.
Based on extensive interviews with the major players, including Marvin Minsky, Herbert Simon, Allen Newell, Raj Reddy, and Patrick Winston, AI is part intellectual history, part business history. Rich with anecdotes about the founders and leaders of the field and their celebrated feuds and intellectual gamesmanship, the book chronicles their dramatic successes ("expert" systems, robotics, "smart" technologies, and even world-class chess playing) and their equally dramatic failures (language processing, learning), and shows how early in the next century researchers hope to teach their computers "common sense," the next necessary breakthrough.
The story of AI is an exhilarating saga of new programs and new hardware, yet it is also the story of a slow but steady acquisition of knowledge about how humans think. Daniel Crevier traces AI's emergence from the fields of philosophy, mathematics, psychology, and neurology, chronicling the development of primitive computing devices and ultimately the creation of a brave new world described chiefly in SOAR, Cyc, EURISKO, among others.
The quest for artificial intelligence raises profound issues about the nature of mind and soul as well as fascinating philosophical questions. Will we humans one day have to share our world with entities smarter than ourselves? And can we rely on these creations to make vital decisions for us - business, scientific, legal, and even moral choices? Crevier discusses these questions with the leaders of AI, and they offer some surprising answers.
Based on extensive interviews with the major players, including Marvin Minsky, Herbert Simon, Allen Newell, Raj Reddy, and Patrick Winston, AI is part intellectual history, part business history. Rich with anecdotes about the founders and leaders of the field and their celebrated feuds and intellectual gamesmanship, the book chronicles their dramatic successes ("expert" systems, robotics, "smart" technologies, and even world-class chess playing) and their equally dramatic failures (language processing, learning), and shows how early in the next century researchers hope to teach their computers "common sense," the next necessary breakthrough.
The story of AI is an exhilarating saga of new programs and new hardware, yet it is also the story of a slow but steady acquisition of knowledge about how humans think. Daniel Crevier traces AI's emergence from the fields of philosophy, mathematics, psychology, and neurology, chronicling the development of primitive computing devices and ultimately the creation of a brave new world described chiefly in SOAR, Cyc, EURISKO, among others.
The quest for artificial intelligence raises profound issues about the nature of mind and soul as well as fascinating philosophical questions. Will we humans one day have to share our world with entities smarter than ourselves? And can we rely on these creations to make vital decisions for us - business, scientific, legal, and even moral choices? Crevier discusses these questions with the leaders of AI, and they offer some surprising answers.
400 pages, Hardcover
First published April 1, 1993
5 people are currently reading
94 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 6 of 6 reviews
November 11, 2016
"If Sue and Jack are talking, and Jim is in the same room, will he understand them?"
"Most likely. Sue, Jack and Jim are familiar forms of the names Susan, John and James, which are characteristic of English-speaking peoples, so most likely Sue, Jack and Jim are native speakers of English. Unless there is a reason not to, and there usually isn't, Sue and Jack will talk in their native language, which Jim is likely to understand."
"What if it is January 2015, and Jim was born in October 2014?"
"No. Jim must be a 2- or a 3-month-old baby, and babies this small do not understand language."
"What if it is January 2015, and Jim was born in October 2010?"
"It depends on Sue's and Jack's ages. If they are adults, then probably not. Adults are interested in topics four-year-old children are unfamiliar with, and use vocabulary far beyond their grasp. However, if Sue and Jack are also four-year-old children, then Jim is likely to understand them very well."
"What if Jim is a dog?"
"No. Dogs do not understand human language other than commands directed at them."
"Suppose Sue, Jack and Jim are adult humans who are native speakers of English and know no other language. If Sue and Jack are talking, and Jim is in the same room, what could make him not understand them?"
"Jim could be asleep, and Sue and Jack could be talking so quietly as to not wake him up."
"If Sue and Jack are talking, and Jim is in the same room, do Sue and Jack know that Jim will understand them?"
"Yes, unless Jim has made himself hidden, for example, by sitting under a table with a long tablecloth."
"And how will this make Jim hidden?"
"Sue and Jack will not see him behind the tablecloth, and if he sits still, they will not hear him, so they will falsely assume that no one else is in the room."
"Why would Jim do that?"
"He could expect Sue and Jack to come to the room, think that they are alone, and discuss matters they would not discuss in front of another person, which, however, Jim wanted to learn."
"What matters?"
"For instance, matters relating to sex or to money."
"How do you know all that?"
"This is something you learn when you grow up as a human being. You are not born knowing these things, but by age 18 you do."
"Can you write down all you know?"
"No. There is just too much and it is unordered."
"Did you learn this at school?"
"No. It is called common sense."
"Do all adult humans know the same things?"
"No. They have different professions. I am a computer programmer, and my brother is a doctor. Last time this came up, I forgot what side of the body the liver is on, and my brother probably hasn't heard of assembly language."
"Can you teach a computer all these things?"
"This book from 1993 says that it took a team of researchers 3 months to make a computer understand the two sentences, "Napoleon died on St. Helena. Wellington was saddened." "
"So was this project successful?"
"The book quotes the head of the project, "No one in 2015 would dream of buying a machine without common sense any more than anyone today would buy a personal computer that couldn't run spreadsheets [or] word processing programs." It is now 2015. In November 1993, the world's fastest computer had maximum performance 124 GFlop/s. In November 2014, it was 34 PFlop/s, more than a quarter million times more. Yet computers still do not have common sense."
"Is there any computer that has something approaching common sense?"
"The closest computer to come to that, IBM's Watson, famously won the game of Jeopardy. It simulated common sense with terabytes of content from "encyclopedias, dictionaries, thesauri, newswire articles, literary works, and so on." An article about Watson says, "the two most likely candidate answers generated by the system for the clue, “In 1594 he took a job as a tax collector in Andalusia,” are “Thoreau” and “Cervantes.” In this case, temporal reasoning is used to rule out Thoreau as he was not alive in 1594, having been born in 1817, whereas Cervantes, the correct answer, was born in 1547 and died in 1616." This is not how a human would approach the problem."
"How would a human approach it?"
"Someone who is not a scholar of 19th century American literature would not know Thoreau's birth year, and someone who is not a scholar of Spanish literature would not know the years of Cervantes's birth and death. However, someone who knows some literature would know that Cervantes was a contemporary of Shakespeare, who lived during the reign of Elizabeth I of England, which was around the year 1600, and that Cervantes's most famous novel is a parody of a knightly romance, so he must have lived close to the time when romances about the adventures of knights-errant were popular, which was the late Middle Ages. Also, someone who knows some geography would know that Andalusia is a province of Spain. A Jeopardy question must be asking about someone famous. Which famous Spanish person who was not a king or a queen (and a king or a queen would not take a job as a tax collector) lived in the late 16th century? It must have been Miguel de Cervantes. No one would ever think of Thoreau. Thoreau is a well-known American writer, and there was no American literature in 1594."
"So is artificial intelligence a fraud?"
"No, not at all. It has been a heroic epic, with researchers trying to make computers millions of times less powerful than today's smartphones, as well as devices that weren't computers, one of which has been described as "a labyrinth of valves, small motors, gears and wires" hooked up to a war-surplus autopilot from a B-24 bomber, behave intelligently. That they did not succeed is not their fault because they did not know what they were up against."
"Why didn't they know it?"
"Because things that seem commonsensical nowadays were not known back then. In the 1956 film "Earth vs. the Flying Saucers" humans capture an alien translation machine, and a character says that by feeding it the dictionary humans would learn the alien language. Noam Chomsky's Syntactic Structures would only revolutionize linguistics the next year."
"Have there been attempts to make a computer do some tasks intelligently even if it cannot approach the intelligence of an educated adult human?"
"Certainly! The Mars Science Laboratory, launched in 2011, can be told to drive to a spot up to 50 meters ahead, and it can use its onboard cameras and planning software to choose its route, avoiding dangerous rocks. There is commercially available speech recognition software, machine translation software, face recognition software. Yet it is all far less than what AI enthusiasts of 50 years ago predicted for 2015."
"Most likely. Sue, Jack and Jim are familiar forms of the names Susan, John and James, which are characteristic of English-speaking peoples, so most likely Sue, Jack and Jim are native speakers of English. Unless there is a reason not to, and there usually isn't, Sue and Jack will talk in their native language, which Jim is likely to understand."
"What if it is January 2015, and Jim was born in October 2014?"
"No. Jim must be a 2- or a 3-month-old baby, and babies this small do not understand language."
"What if it is January 2015, and Jim was born in October 2010?"
"It depends on Sue's and Jack's ages. If they are adults, then probably not. Adults are interested in topics four-year-old children are unfamiliar with, and use vocabulary far beyond their grasp. However, if Sue and Jack are also four-year-old children, then Jim is likely to understand them very well."
"What if Jim is a dog?"
"No. Dogs do not understand human language other than commands directed at them."
"Suppose Sue, Jack and Jim are adult humans who are native speakers of English and know no other language. If Sue and Jack are talking, and Jim is in the same room, what could make him not understand them?"
"Jim could be asleep, and Sue and Jack could be talking so quietly as to not wake him up."
"If Sue and Jack are talking, and Jim is in the same room, do Sue and Jack know that Jim will understand them?"
"Yes, unless Jim has made himself hidden, for example, by sitting under a table with a long tablecloth."
"And how will this make Jim hidden?"
"Sue and Jack will not see him behind the tablecloth, and if he sits still, they will not hear him, so they will falsely assume that no one else is in the room."
"Why would Jim do that?"
"He could expect Sue and Jack to come to the room, think that they are alone, and discuss matters they would not discuss in front of another person, which, however, Jim wanted to learn."
"What matters?"
"For instance, matters relating to sex or to money."
"How do you know all that?"
"This is something you learn when you grow up as a human being. You are not born knowing these things, but by age 18 you do."
"Can you write down all you know?"
"No. There is just too much and it is unordered."
"Did you learn this at school?"
"No. It is called common sense."
"Do all adult humans know the same things?"
"No. They have different professions. I am a computer programmer, and my brother is a doctor. Last time this came up, I forgot what side of the body the liver is on, and my brother probably hasn't heard of assembly language."
"Can you teach a computer all these things?"
"This book from 1993 says that it took a team of researchers 3 months to make a computer understand the two sentences, "Napoleon died on St. Helena. Wellington was saddened." "
"So was this project successful?"
"The book quotes the head of the project, "No one in 2015 would dream of buying a machine without common sense any more than anyone today would buy a personal computer that couldn't run spreadsheets [or] word processing programs." It is now 2015. In November 1993, the world's fastest computer had maximum performance 124 GFlop/s. In November 2014, it was 34 PFlop/s, more than a quarter million times more. Yet computers still do not have common sense."
"Is there any computer that has something approaching common sense?"
"The closest computer to come to that, IBM's Watson, famously won the game of Jeopardy. It simulated common sense with terabytes of content from "encyclopedias, dictionaries, thesauri, newswire articles, literary works, and so on." An article about Watson says, "the two most likely candidate answers generated by the system for the clue, “In 1594 he took a job as a tax collector in Andalusia,” are “Thoreau” and “Cervantes.” In this case, temporal reasoning is used to rule out Thoreau as he was not alive in 1594, having been born in 1817, whereas Cervantes, the correct answer, was born in 1547 and died in 1616." This is not how a human would approach the problem."
"How would a human approach it?"
"Someone who is not a scholar of 19th century American literature would not know Thoreau's birth year, and someone who is not a scholar of Spanish literature would not know the years of Cervantes's birth and death. However, someone who knows some literature would know that Cervantes was a contemporary of Shakespeare, who lived during the reign of Elizabeth I of England, which was around the year 1600, and that Cervantes's most famous novel is a parody of a knightly romance, so he must have lived close to the time when romances about the adventures of knights-errant were popular, which was the late Middle Ages. Also, someone who knows some geography would know that Andalusia is a province of Spain. A Jeopardy question must be asking about someone famous. Which famous Spanish person who was not a king or a queen (and a king or a queen would not take a job as a tax collector) lived in the late 16th century? It must have been Miguel de Cervantes. No one would ever think of Thoreau. Thoreau is a well-known American writer, and there was no American literature in 1594."
"So is artificial intelligence a fraud?"
"No, not at all. It has been a heroic epic, with researchers trying to make computers millions of times less powerful than today's smartphones, as well as devices that weren't computers, one of which has been described as "a labyrinth of valves, small motors, gears and wires" hooked up to a war-surplus autopilot from a B-24 bomber, behave intelligently. That they did not succeed is not their fault because they did not know what they were up against."
"Why didn't they know it?"
"Because things that seem commonsensical nowadays were not known back then. In the 1956 film "Earth vs. the Flying Saucers" humans capture an alien translation machine, and a character says that by feeding it the dictionary humans would learn the alien language. Noam Chomsky's Syntactic Structures would only revolutionize linguistics the next year."
"Have there been attempts to make a computer do some tasks intelligently even if it cannot approach the intelligence of an educated adult human?"
"Certainly! The Mars Science Laboratory, launched in 2011, can be told to drive to a spot up to 50 meters ahead, and it can use its onboard cameras and planning software to choose its route, avoiding dangerous rocks. There is commercially available speech recognition software, machine translation software, face recognition software. Yet it is all far less than what AI enthusiasts of 50 years ago predicted for 2015."
April 19, 2023
I read this for school, I had to write a paper on it. It was a good book actually. It talked about chess in it but was not overly focused on it. It talked about AI and machine learning in general at a more wide view and less in depth. Not that that’s a problem at all. It was good and enjoyable.
4/5
4/5
November 24, 2025
Artificial intelligence up to the 90s
November 21, 2010
We're all doomed! The book concludes that we are probably going to design the computer intelligences that will replace the human race. It seems like these guys don't learn from their past punctured hubris. After describing all the exaggerated hopes of early AI in the 60s, 70s and 80s (like a student being given a summer project to 'solve the vision problem', or the 1980s Japanese project to make an AI capable of holding a casual conversation), and then saying that AI people are more careful about making crazy predictions these days, he goes on to make a whole bunch of them in the conclusion of the book. Although much of early AI was funded by DARPA, the author is fairly critical of the use of AI in military kit, particular nuclear weapons.
You might like it more if you hold the strong AI position. I'm just a disillusioned programmer, who realises that sadly he can't be god, and that really 'Artificial Intelligence' is just 'algorithms substituting intelligence'. But maybe you can evolve things smarter than us.
You might like it more if you hold the strong AI position. I'm just a disillusioned programmer, who realises that sadly he can't be god, and that really 'Artificial Intelligence' is just 'algorithms substituting intelligence'. But maybe you can evolve things smarter than us.
January 7, 2013
Well-written, thoughtful, if a little dated (the last third, when it mentions the advent of email and virtual reality, is particularly out-of-date). My favorite section was the middle third, discussing the rise of of expert systems. There is also some good stuff re: Hubert Dreyfus' battles with MIT's AI researchers. I love that academic vs. academic stuff, it's almost trashy -- well, trashy as you can get in a history of AI.
July 28, 2015
This is my first AI book and I really enjoyed it. I particularly liked the history of computer hardware and software development, showing the insights and challenges and setbacks on the way to the AI and expert systems we enjoy now. The book is dated but is worth the read, especially as your first book - introducing you to the subject.
Displaying 1 - 6 of 6 reviews