What do you think?
Rate this book
A Brief Guide to Artificial Intelligence
Artificial intelligence (AI) has now mastered tasks that until recently could be performed only by humans. These tasks include cancer diagnosis, drug design, object recognition, speech recognition, and playing chess, backgammon and Go, which AI systems perform at superhuman levels. This richly illustrated book is a brief but comprehensive overview (without equations) of current AI systems, how they work, their applications, and their limitations. After surveying the impressive capabilities of AI systems in certain domains, the limited ability of AI to perform tasks that humans find trivial is discussed. Finally, the question of whether AI systems can be intelligent is considered, along with the controversial issue of machine consciousness. Written in an informal style, with a comprehensive Glossary and a list of Further Readings, this book is an ideal introduction to the rapidly evolving field of AI.
Kindle Edition
First published April 1, 2020
26 people want to read
About the author
James V. Stone
21 books34 followersHonorary Associate Professor, University of Sheffield, England.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 3 of 3 reviews
April 18, 2020
Some brief guides miss the point and in reality go on for rather a long time, but this one very much does what it says on the tin. Readable in an hour, it gives us the basics of what modern artificial intelligence is and how it works. To achieve such brevity, James Stone has cut away much of the support mechanism of popular science - so we get very little background and historical context, limited storytelling and not much about applications. Instead, the focus is laser-sharp on delivering the basics of how neural networks work, where modern AI is successful and where it has a long way to go.
Broadly, Stone identifies two areas where AI triumphs - image recognition and playing games. If I'm honest, perhaps he is a little generous on the first of these - as he points out himself, tiny elements undetectable to the human eye can be sufficient to totally change what an image is recognised to be, and most of the techniques (he takes us through different types of machine learning) require a very large amount of training material where (again as Stone acknowledges) humans can often learn to recognise something with just a handful of examples. The AI technology is on firmer ground with its game playing - in chess, go, backgammon and a range of computer games, AI is now unbeatable, and teaching humans new approaches. What perhaps Stone doesn't emphasise enough is how much this demonstrates that AI's real success is in non-real world applications where the rules are clear and relatively simple, even if they haven't been specified to the system. In the real world, things are often more messy, again something that is acknowledged in, for example, being far less optimistic about self-driving cars than is often the case.
Stone gives us a good simple introduction to neural networks, back propagation, four different kinds of machine learning (from semi-supervised to reinforcement) and the dangers of overfitting, making it clear that even current AI is doing significantly more than curve fitting. The book (probably sensibly) brushes aside fears about AI might take over the world and displace humans as somewhat far fetched at the moment.
My only real disappointment was at the end of the book, where Stone tells us true AI is likely to emerge relatively soon, given it took just millions of years for flying to develop in the natural world, but just 66 to get from the Wright brothers to the Moon landing, citing acceleration of advancement that could be experimental. There are two problems with this. One is that the proper modern comparison with the Wright brothers is not space travel but commercial flights. These did indeed get much faster. The Wright brothers had an airspeed in the tens of miles per hour. Within 73 years, passengers could fly at 1,350 mph. But Concorde has now been withdrawn for nearly 20 years - instead of speeding up, we fly at less than half the speed. Even if you do take the leap from Wright brothers to Apollo - yes, that only took 66 years. But in over 50 years since we have done nothing comparable in space, let alone gone further and faster. Just because there was acceleration in the past does not mean it will continue.
I give the guide four stars for its solid, approachable and extremely brief introduction to the modern field, but three stars for lacking context and effective narrative. While approachable and doing the job admirably, it's not outstandingly particularly to read without that contextual material. Even so, for the right purpose it's an excellent little book.
Broadly, Stone identifies two areas where AI triumphs - image recognition and playing games. If I'm honest, perhaps he is a little generous on the first of these - as he points out himself, tiny elements undetectable to the human eye can be sufficient to totally change what an image is recognised to be, and most of the techniques (he takes us through different types of machine learning) require a very large amount of training material where (again as Stone acknowledges) humans can often learn to recognise something with just a handful of examples. The AI technology is on firmer ground with its game playing - in chess, go, backgammon and a range of computer games, AI is now unbeatable, and teaching humans new approaches. What perhaps Stone doesn't emphasise enough is how much this demonstrates that AI's real success is in non-real world applications where the rules are clear and relatively simple, even if they haven't been specified to the system. In the real world, things are often more messy, again something that is acknowledged in, for example, being far less optimistic about self-driving cars than is often the case.
Stone gives us a good simple introduction to neural networks, back propagation, four different kinds of machine learning (from semi-supervised to reinforcement) and the dangers of overfitting, making it clear that even current AI is doing significantly more than curve fitting. The book (probably sensibly) brushes aside fears about AI might take over the world and displace humans as somewhat far fetched at the moment.
My only real disappointment was at the end of the book, where Stone tells us true AI is likely to emerge relatively soon, given it took just millions of years for flying to develop in the natural world, but just 66 to get from the Wright brothers to the Moon landing, citing acceleration of advancement that could be experimental. There are two problems with this. One is that the proper modern comparison with the Wright brothers is not space travel but commercial flights. These did indeed get much faster. The Wright brothers had an airspeed in the tens of miles per hour. Within 73 years, passengers could fly at 1,350 mph. But Concorde has now been withdrawn for nearly 20 years - instead of speeding up, we fly at less than half the speed. Even if you do take the leap from Wright brothers to Apollo - yes, that only took 66 years. But in over 50 years since we have done nothing comparable in space, let alone gone further and faster. Just because there was acceleration in the past does not mean it will continue.
I give the guide four stars for its solid, approachable and extremely brief introduction to the modern field, but three stars for lacking context and effective narrative. While approachable and doing the job admirably, it's not outstandingly particularly to read without that contextual material. Even so, for the right purpose it's an excellent little book.
Read
March 31, 2020"An excellent short primer to quickly get the uninitiated grounded in the field of artificial intelligence."
Professor Eric Topol, MD, Executive Vice-President, Scripps Research, USA.
"There are very few people that have both a technical understanding of recent achievements in artificial intelligence, and a deep appreciation of the nature of human intelligence. James Stone is one of those people. Fortunately for us, he has distilled much of his understanding into this book, which doesn't sacrifice accuracy for brevity, and which summarises recent achievements, as well as the challenges that lie ahead. I'm often asked to recommend a short introduction to artificial intelligence; this book will do the job perfectly."
Neil Lawrence, DeepMind Professor of Machine Learning, University of Cambridge, and Senior AI Fellow at the Turing Institute.
"This is the book you need if you want to understand the basics of AI, including its triumphs and limitations. Stone takes the reader on an authoritative whirlwind tour, and delves into just enough technical detail to give an overview of key concepts, such as neural networks, deepfakes, and overfitting. An essential and timely summary of modern AI. "
James Marshall, Professor of Computational Biology, University of Sheffield, and Co-Founder of Opteran Technologies.
Professor Eric Topol, MD, Executive Vice-President, Scripps Research, USA.
"There are very few people that have both a technical understanding of recent achievements in artificial intelligence, and a deep appreciation of the nature of human intelligence. James Stone is one of those people. Fortunately for us, he has distilled much of his understanding into this book, which doesn't sacrifice accuracy for brevity, and which summarises recent achievements, as well as the challenges that lie ahead. I'm often asked to recommend a short introduction to artificial intelligence; this book will do the job perfectly."
Neil Lawrence, DeepMind Professor of Machine Learning, University of Cambridge, and Senior AI Fellow at the Turing Institute.
"This is the book you need if you want to understand the basics of AI, including its triumphs and limitations. Stone takes the reader on an authoritative whirlwind tour, and delves into just enough technical detail to give an overview of key concepts, such as neural networks, deepfakes, and overfitting. An essential and timely summary of modern AI. "
James Marshall, Professor of Computational Biology, University of Sheffield, and Co-Founder of Opteran Technologies.
May 13, 2020
Another great work from James Stone.
A very precise, concise and up-to-date introduction to AI. I am always amazing how the author is able to provide so much content, explained with great clarity, in a very small amount of pages. I really enjoyed reading it!
A very precise, concise and up-to-date introduction to AI. I am always amazing how the author is able to provide so much content, explained with great clarity, in a very small amount of pages. I really enjoyed reading it!
Displaying 1 - 3 of 3 reviews