A Brief Guide to Artificial Intelligence
About this book
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.
Published 1 April 2020
Paperback ISBN: 9781916279117
Kindle ISBN: 9781916279124
Download Chapter 1 (PDF, 4.9MB)
Available from:
Contents
Preface
1. What is artificial intelligence?
1.1 Introduction
1.2 AI, AGI and other euphemisms
1.3 So, what is AI?
1.4 Almost human memory
1.5 Interpretability
1.6 Why now?
2. What AI can, and cannot, do
2.1 Introduction
2.2 Visual recognition
2.3 Deepfakes
2.4 The player of games
2.5 Skilled control
2.6 AI in medicine
2.7 Language
2.8 Is AI biased?
2.9 What AI cannot do
3. How AI works
3.1 Introduction
3.2 Supervised learning
3.3 Convolutional neural networks
3.4 The backprop algorithm
3.5 Semi-supervised learning
3.6 Self-supervised learning
3.7 Unsupervised learning
3.8 Reinforcement learning
3.9 Over-fitting
3.10 Is AI just curve-fitting?
4. The future of AI
4.1 Introduction
4.2 Time enough for AI?
4.3 Another revolution?
4.4 Machine consciousness
Further reading
Glossary
References
Index
Reviews
"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.
"Some brief guides miss the point, and go on for rather a long time, but this one very much does what it says on the tin... 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."
Brian Clegg, author of The Quantum Age