What do you think?
Rate this book
Machines That Think: The Future of Artificial Intelligence
A scientist who has spent a career developing Artificial Intelligence takes a realistic look at the technological challenges and assesses the likely effect of AI on the future.
How will Artificial Intelligence impact our lives? Toby Walsh, one of the leading AI researchers in the world, takes a critical look at the impact Artificial Intelligence is set to have on our lives.
Based on a deep understanding of the technology, Walsh describes where Artificial Intelligence is today, and where it will take us.
• Will automation take away most of our jobs?
• Is a "technological singularity" near?
• What is the chance that robots will take over?
• How do we best prepare for this future?
The author concludes that, if we plan well, thinking machines could be our greatest legacy, the last invention human beings will ever need to make.
How will Artificial Intelligence impact our lives? Toby Walsh, one of the leading AI researchers in the world, takes a critical look at the impact Artificial Intelligence is set to have on our lives.
Based on a deep understanding of the technology, Walsh describes where Artificial Intelligence is today, and where it will take us.
• Will automation take away most of our jobs?
• Is a "technological singularity" near?
• What is the chance that robots will take over?
• How do we best prepare for this future?
The author concludes that, if we plan well, thinking machines could be our greatest legacy, the last invention human beings will ever need to make.
352 pages, Paperback
Published February 13, 2018
20 people are currently reading
177 people want to read
About the author
Toby Walsh
33 books34 followersToby Walsh is one of the world’s leading researchers in Artificial Intelligence. He is a Professor of Artificial Intelligence at the University of New South Wales and leads a research group at Data61, Australia’s Centre of Excellence for ICT Research. He has been elected a fellow of the Association for the Advancement of AI for his contributions to AI research, and has won the prestigious Humboldt research award. He has previously held research positions in England, Scotland, France, Germany, Italy, Ireland and Sweden.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 13 of 13 reviews
May 28, 2020
So, some of the others who were reading this book with me argued that we should let the AIs control themselves. So if we ever have a Matrix scenario, I know who to blame.
Review first published on my blog: https://agencyofbooksandspies.blogspo...
Review first published on my blog: https://agencyofbooksandspies.blogspo...
February 23, 2019
Thoroughly enjoyed the book, but was hoping it would have been a bit more technical. I enjoyed the historical survey of AI that was presented followed by the topics touching on the promises of AI, the ethical questions and the potential technical limitations which could limit AI and not result in a robot ruled world. I read nearly all of the book on a flight down and back to ATL, so it was a quick read.
December 6, 2024
A very good overview of the subject geared towards the layman and non-AI specialist
Any review of this book would have to start out with the intended audience of this book. The author is pretty explicit in defining this. On the last page of the prologue he writes "This book is intended for the interested but not specialist reader". However, Dr. Walsh seems to imply that the definition of the "specialist reader" is basically anyone with a non-technical background (i.e., electrical engineers and programmers as well as Al specialists) but the fact of the matter, at least in this reviewer's opinion, the real definition of "specialist" is anyone with a deep level of Al knowledge. In this reviewer's opinion even those with relatively technical backgrounds in fields like programming and electrical engineering but with limited Al experience would find Dr. Walsh's views interesting.
The book itself is broken out into three broad parts, the past, present and future of Al. The first section provides a historical overview of Al and factors important to it. There are discussions of how the development of Boolean/symbolic logic, early cybernetic theory and the development of electrical hardware all played an interconnected role to get us to the point where we are today. The chapter hence fulfills the important need to provide historical perspective. As Cicero once said "A man who does not know what has happened before he was born is an ignoramus" and as Aristotle once said "A man without perspective has no value". This chapter fulfills this role well.
The next chapter discusses the current state of Al. This includes a discussion of current schools of thought, limits of Al and its current state. He provides numerous interesting points (and counter-points) here. For example, he discusses the power of Bayesian networks, on the one hand, as well as Roger Penrose's view that sophisticated Al will never be achieved due to the limits of logic (defined in terms of Boolean/symbolic logic terms). (For those interested in Dr. Penrose's view this reviewer very highly recommends his "The Emperor's New Mind"). He also discusses the current state of artificial intelligence in fields such as autonomous driving, among others.
In the last chapter he discusses his views on Al's future. He makes the relatively straight forward observation that advanced Al, even if it is not up to the par of a "thinking machine" in lines with human thought, will be, using that over worn cliche, seriously "disruptive". There will be both positive benefits (i.e., humans will no longer have to work in dangerous occupations such as mining) as well as negative (i.e., massive unemployment). He also posits that we will probably not be able to reach the point of the "singularity", as put forth by Ray Kurzweil in his book "The Singularity Is Near: When Humans Transcend Biology". He concludes with 10 predictions regarding the impact of Al in the future (specifically the year 2050). They are:
a) Autonomous driving will displace human driving
b) Health will improve as various Fitbit like devices will provide perpetual real time data to Al "doctors" who can provide input back to their users.
c) Virtual technology will advance to the point where movies and computer games fuse.
d) Al will play a significant role in HR style hiring and firing decisions.
e) There will be massive interactivity between humans and just about everything around them.
The "internet of things" will come to total functioning fruition hence people will be able to interact via even simple devices such as coffee makers and toasters via voice commands. In this sub-section of the book, like the others on his other predictions, he points out both the benefits and negatives of this. The positive being the obvious increases in convenience and productivity and the negative the obvious loss of privacy.
f) Al will reach the point where it can autonomously commit serious electronic crimes, including bank robbery.
g) Electronic device's dexterity will reach the point where they will enable Al to actually have a "body" that is quite functional. This will reach the point where robots will finally be able to beat humans in soccer's World Cup in much the same way they defeated Gary Kasparov in chess in the mid-1990s.
h) Autonomous planes, ships and trains (with no human controllers) will become the norm.
i) Journalism will reach the point where no humans will be needed (at least in its presentation and most of its writing).
j) Al will enable us to live on electronically, in terms of chatbots and programs that mimic our personalities.
In this reviewer's opinion this is the weakest chapter of the book as most of the predictions are pretty self evident, even to the non-specialist. However, Dr. Walsh does present us with both the positive and negatives associated with each and many of these views are quite interesting.
The book also has very interesting factoids throughout that are quite interesting. For example, that Al has reached the point where Deep Fritz, a commercially available program, was able to beat the human that took over Gary Kasparov's title of human world chess champion (as rated by the World Chess Federation) on a regular stand along PC only about 10 years after Deep Blue was able to defeat Gary Kasparov. This is of extreme interest as the so-called "computer" that defeated Mr. Kasparov was actually a series of three mainframes working in parallel and had many high-level programmers handling it (i.e., fine tuning the program, hardware, etc.). That a version of this program running on smart phone has recently reached the same chess rating as Gary Kasparov's when he was the world's chess champion. All and all very interesting.
Despite all these positives the book does have a few negatives of note. One is that he does not elaborate more, in this reviewer's opinion, on the debate between proponents who posit that Al will reach human like intelligence and those, like Dr. Penrose, who disagree. Dr. Walsh should have expounded much more on this than he has in this book, especially considering the immense future implications. Another weakness is his lack of discussion between the development of electronic machine's physical dexterity and Al. This, too, has serious implications as it can provide Al with the "body" it needs to disrupt so much human activity, especially in the labour markets. (For a good book on this subject this reviewer highly recommends Martin Ford's "Rise of the Robots").
Lastly, there is not enough discussion of how to counter the negative impacts of automation's negative impact on the labour market (Dr Walsh explicitly states that automation will probably put, conservatively, 25% of the workforce out of work). Dr. Walsh believes, like Elon Musk and the Austrian school of economists, that this can best be handled through a "minimum income" scheme whereby people are just given money.
This is despite the fact that, even as Dr. Walsh himself admits, this will cost a fortune. Dr. Walsh estimates that the provision of an "income" of just $18,000, per person, will cost about $3.2 trillion in the US (the US's GDP is currently about $19 trillion). Plus, more importantly, he just assumes that the extremely wealthy (and those who will be the primary beneficiaries of the up and coming revolutionary automation) will just hand over this money to the government without a fight. In this reviewer's opinion, considering what has actually happened historically, many will prefer to live under a Pinochet or Franco and keep their fortunes than a parliamentary democratic system that will dispossess them of much of their income. Considering the fact that these problems are among the most important in the book this is more than a minor insight.
In short, despite the aforementioned weaknesses, this book is highly recommended for both a lay audience and even a technical one (i.e., electrical engineers and programmers who may not necessarily be Al experts). Four stars.
Any review of this book would have to start out with the intended audience of this book. The author is pretty explicit in defining this. On the last page of the prologue he writes "This book is intended for the interested but not specialist reader". However, Dr. Walsh seems to imply that the definition of the "specialist reader" is basically anyone with a non-technical background (i.e., electrical engineers and programmers as well as Al specialists) but the fact of the matter, at least in this reviewer's opinion, the real definition of "specialist" is anyone with a deep level of Al knowledge. In this reviewer's opinion even those with relatively technical backgrounds in fields like programming and electrical engineering but with limited Al experience would find Dr. Walsh's views interesting.
The book itself is broken out into three broad parts, the past, present and future of Al. The first section provides a historical overview of Al and factors important to it. There are discussions of how the development of Boolean/symbolic logic, early cybernetic theory and the development of electrical hardware all played an interconnected role to get us to the point where we are today. The chapter hence fulfills the important need to provide historical perspective. As Cicero once said "A man who does not know what has happened before he was born is an ignoramus" and as Aristotle once said "A man without perspective has no value". This chapter fulfills this role well.
The next chapter discusses the current state of Al. This includes a discussion of current schools of thought, limits of Al and its current state. He provides numerous interesting points (and counter-points) here. For example, he discusses the power of Bayesian networks, on the one hand, as well as Roger Penrose's view that sophisticated Al will never be achieved due to the limits of logic (defined in terms of Boolean/symbolic logic terms). (For those interested in Dr. Penrose's view this reviewer very highly recommends his "The Emperor's New Mind"). He also discusses the current state of artificial intelligence in fields such as autonomous driving, among others.
In the last chapter he discusses his views on Al's future. He makes the relatively straight forward observation that advanced Al, even if it is not up to the par of a "thinking machine" in lines with human thought, will be, using that over worn cliche, seriously "disruptive". There will be both positive benefits (i.e., humans will no longer have to work in dangerous occupations such as mining) as well as negative (i.e., massive unemployment). He also posits that we will probably not be able to reach the point of the "singularity", as put forth by Ray Kurzweil in his book "The Singularity Is Near: When Humans Transcend Biology". He concludes with 10 predictions regarding the impact of Al in the future (specifically the year 2050). They are:
a) Autonomous driving will displace human driving
b) Health will improve as various Fitbit like devices will provide perpetual real time data to Al "doctors" who can provide input back to their users.
c) Virtual technology will advance to the point where movies and computer games fuse.
d) Al will play a significant role in HR style hiring and firing decisions.
e) There will be massive interactivity between humans and just about everything around them.
The "internet of things" will come to total functioning fruition hence people will be able to interact via even simple devices such as coffee makers and toasters via voice commands. In this sub-section of the book, like the others on his other predictions, he points out both the benefits and negatives of this. The positive being the obvious increases in convenience and productivity and the negative the obvious loss of privacy.
f) Al will reach the point where it can autonomously commit serious electronic crimes, including bank robbery.
g) Electronic device's dexterity will reach the point where they will enable Al to actually have a "body" that is quite functional. This will reach the point where robots will finally be able to beat humans in soccer's World Cup in much the same way they defeated Gary Kasparov in chess in the mid-1990s.
h) Autonomous planes, ships and trains (with no human controllers) will become the norm.
i) Journalism will reach the point where no humans will be needed (at least in its presentation and most of its writing).
j) Al will enable us to live on electronically, in terms of chatbots and programs that mimic our personalities.
In this reviewer's opinion this is the weakest chapter of the book as most of the predictions are pretty self evident, even to the non-specialist. However, Dr. Walsh does present us with both the positive and negatives associated with each and many of these views are quite interesting.
The book also has very interesting factoids throughout that are quite interesting. For example, that Al has reached the point where Deep Fritz, a commercially available program, was able to beat the human that took over Gary Kasparov's title of human world chess champion (as rated by the World Chess Federation) on a regular stand along PC only about 10 years after Deep Blue was able to defeat Gary Kasparov. This is of extreme interest as the so-called "computer" that defeated Mr. Kasparov was actually a series of three mainframes working in parallel and had many high-level programmers handling it (i.e., fine tuning the program, hardware, etc.). That a version of this program running on smart phone has recently reached the same chess rating as Gary Kasparov's when he was the world's chess champion. All and all very interesting.
Despite all these positives the book does have a few negatives of note. One is that he does not elaborate more, in this reviewer's opinion, on the debate between proponents who posit that Al will reach human like intelligence and those, like Dr. Penrose, who disagree. Dr. Walsh should have expounded much more on this than he has in this book, especially considering the immense future implications. Another weakness is his lack of discussion between the development of electronic machine's physical dexterity and Al. This, too, has serious implications as it can provide Al with the "body" it needs to disrupt so much human activity, especially in the labour markets. (For a good book on this subject this reviewer highly recommends Martin Ford's "Rise of the Robots").
Lastly, there is not enough discussion of how to counter the negative impacts of automation's negative impact on the labour market (Dr Walsh explicitly states that automation will probably put, conservatively, 25% of the workforce out of work). Dr. Walsh believes, like Elon Musk and the Austrian school of economists, that this can best be handled through a "minimum income" scheme whereby people are just given money.
This is despite the fact that, even as Dr. Walsh himself admits, this will cost a fortune. Dr. Walsh estimates that the provision of an "income" of just $18,000, per person, will cost about $3.2 trillion in the US (the US's GDP is currently about $19 trillion). Plus, more importantly, he just assumes that the extremely wealthy (and those who will be the primary beneficiaries of the up and coming revolutionary automation) will just hand over this money to the government without a fight. In this reviewer's opinion, considering what has actually happened historically, many will prefer to live under a Pinochet or Franco and keep their fortunes than a parliamentary democratic system that will dispossess them of much of their income. Considering the fact that these problems are among the most important in the book this is more than a minor insight.
In short, despite the aforementioned weaknesses, this book is highly recommended for both a lay audience and even a technical one (i.e., electrical engineers and programmers who may not necessarily be Al experts). Four stars.
January 10, 2019
At once sobering and hopeful, Walsh gives a very good overall view of the recent past of AI as well as the future concerns. There are some 'Terminator' fears that are not ungrounded; but also the optimism of better medicine, safer driving, personalized continuing education, and other benefits we cannot yet imagine.
It is impossible to predict the future, but it will get here regardless. From men with red flags walking in front of the new 'horseless carriage' back in the day, to a ban on anti-personnel mines and blinding weapons on the battlefield today, Walsh reminds us that humanity is often locked in a teeter-totter of reacting to new inventions. The time to discuss the ethics of self-driving cars, Siri and Alexa, facial recognition in public places, AI genome mapping and manipulation, and all double-edged technologies is, optimally, before they are introduced. Not a bad book to look at, and then to talk to others about; the conversation is important.
It is impossible to predict the future, but it will get here regardless. From men with red flags walking in front of the new 'horseless carriage' back in the day, to a ban on anti-personnel mines and blinding weapons on the battlefield today, Walsh reminds us that humanity is often locked in a teeter-totter of reacting to new inventions. The time to discuss the ethics of self-driving cars, Siri and Alexa, facial recognition in public places, AI genome mapping and manipulation, and all double-edged technologies is, optimally, before they are introduced. Not a bad book to look at, and then to talk to others about; the conversation is important.
March 21, 2019
A fast-paced read and overall an accessible introduction to A.I. science and how we got to where we are today in terms of machine learning and applying algorithms to harness big data. As someone who took calc and physics in high school but hasn’t formally studied math, science or programming since then, I could easily understand the discussion and found the historical context enlightening. But to his detriment, Walsh relies heavily on a small set of examples — the A.I. AlphaGo defeating the world’s best human player in 2016 is one of his go-tos — that get rather repetitive by the end of the book, and he often avoids going into more detail just when you’d like him to delve deeper — especially when making lists and summarizing conclusions without exploring obvious counterpoints.
March 6, 2018
A mandatory read
Predicting the future is tough but this author shows the projected capabilities of computing technologies and how disruptive they can be. You owe it to yourself to be aware.
Predicting the future is tough but this author shows the projected capabilities of computing technologies and how disruptive they can be. You owe it to yourself to be aware.
February 20, 2020
I'm in the midst of researching a paper about Autonomous Weapons and came upon this book. It was a great addition to my research and a good read for anyone interested in AI.
Whereas Bostrom took a very philosophical approach to what he termed as "superintelligence", which is not necessarily machine intelligence but could also be the result of genetic engineering, Walsh focuses on machine AI and takes a more practical approach to his subject. These two books compliment each other quite nicely in fact.
Walsh starts by giving an overview of the history of AI, making a rather convincing argument that we can place the roots of today's AI in Greece with Aristotle's development of logic, without which we would have no computers at all. Very cool to think about if you ask me. He then turns to an overview of the state of AI today, which has developed much faster than the laws governing it, and concludes by making predictions about what we will see by 2050.
Walsh develops his ideas in a way that's easy to follow and his writing is very reader friendly, making it easy and fun for the reader to engage with his ideas.
It's easy when looking at how advanced AI is today and its potential to seriously impact society to jump to the conclusion that AI is scary or inherently bad in some way. Walsh, however, makes an important point by stating that technology like AI is morally neutral, it's humans who make it good or bad through their application. This means, as Wash points out, that as a society we need to have conversations about this technology and how we intend to use it, rather than just leaving it up to a few people to decide what's best for the rest of us.
All in all, Wash makes a good contribution to the present conversations surrounding AI, and with the technology developing as fast as it is, this is a very worthwhile book.
4/5 Stars
Whereas Bostrom took a very philosophical approach to what he termed as "superintelligence", which is not necessarily machine intelligence but could also be the result of genetic engineering, Walsh focuses on machine AI and takes a more practical approach to his subject. These two books compliment each other quite nicely in fact.
Walsh starts by giving an overview of the history of AI, making a rather convincing argument that we can place the roots of today's AI in Greece with Aristotle's development of logic, without which we would have no computers at all. Very cool to think about if you ask me. He then turns to an overview of the state of AI today, which has developed much faster than the laws governing it, and concludes by making predictions about what we will see by 2050.
Walsh develops his ideas in a way that's easy to follow and his writing is very reader friendly, making it easy and fun for the reader to engage with his ideas.
It's easy when looking at how advanced AI is today and its potential to seriously impact society to jump to the conclusion that AI is scary or inherently bad in some way. Walsh, however, makes an important point by stating that technology like AI is morally neutral, it's humans who make it good or bad through their application. This means, as Wash points out, that as a society we need to have conversations about this technology and how we intend to use it, rather than just leaving it up to a few people to decide what's best for the rest of us.
All in all, Wash makes a good contribution to the present conversations surrounding AI, and with the technology developing as fast as it is, this is a very worthwhile book.
4/5 Stars
March 2, 2019
This is a basic book written in a pedestrian style. I learned nothing. It was interesting only as a contrast to Nick Bostrum's greatly superior "Superintelligence." Bostrum takes a bleak view of the potential dangers of superintelligence and seems certain that it is coming soon whether we like it or not. Walsh emphasizes how hard artifical intelligence is and how great the remaining obstacles are. He has serious doubts that we will ever attain superintelligence. Admitting that I am only modestly knowledgeable in this area, I lean toward the Bostrum perspective.
August 17, 2018
Picked this up at the new non fiction area of the library. I learned a great deal about the history of mathematics and might be able to convey some of this to my son the math professor. I particularly enjoyed the end of the book when the author discussed technological change and the future. There are lessons and predictions in the last few chapters.
February 17, 2019
I was originally going to give 3 stars as i was just feeling that I am taking a journey of all the innovations in human history and reading manuals from various innovators work. But then I decided to add 4th star merely for the last quarter of book when author actually starts talking about future and his perception about the future.
December 27, 2019
An interesting book on the past and future of AI. Not too deep, though.
October 20, 2020
A book that provides a decent introduction into AI. I think it could go in depth a bit more on certain subjects, however it was plenty for the average person.
July 20, 2018
This book gives you a really good idea about AI, what it can do and how it’ll change our society. Makes you want to learn more about he individual topics.
Displaying 1 - 13 of 13 reviews