This is an excellent book. It explains AI well and provides more information on the subject. I'd definitely read it again.

AI is hotter than glowing coals right now and I was attracted to this book by the expertise of the author (Tim Rocktäschel is a Professor of Artificial Intelligence at University College London and also a researcher at Google) as well as the “10 things you should know” theme. AI is amazing but also perhaps over-hyped and I was hoping for some debunking of common assumptions as well as insight into what technology can do.

The book is written for the general reader, and the author says that it does not require computing or mathematical knowledge.

The first chapter asks the question: what is AI? An excellent question and I was somewhat disappointed. The author distinguishes between domain-specific and general-purpose AI, but for me did not explain what it is.

Next comes a chapter on neural networks. This does a better job than the first in answering the matter of what AI is. Using the example of image recognition, we learn that it is about training algorithms on data; and that processing the data is compute-intensive and ideal for processing using the same kind of hardware as is used for fast computer graphics (GPUs). I have questions though. What if the data is bad? What kinds of mistakes is this approach prone to? Maybe this will come later.

Chapter 3 asserts that superhuman AI is attainable. It is all a bit speculative though. Rocktäschel touches on but does not fully engage with questions like what is consciousness. To me, the question needs breaking down. Is it not the case that AI is already superhuman in some respects, and subhuman in others? There is also the matter of how much AI can improve itself. The author believes that it can, or will be able to, but might there be limitations on that?

Next up is games. AI is very good at chess, we learn, and pretty good at Go. I think there is a lot more to say on this subject, especially when it comes to games where information is hidden, such as most card games, and in which players can bluff and gamble.

Chapter 5 is about the benefits of more data. Chapter 6 is about chatbots, and begins to look at the problem of bad data but has more questions than answers.

Chapter 7 diverges into the matter of AI making scientific discoveries, where the author states his belief that this is beginning to happen. To my mind this is something that belonged with chapter 3, as it touches on how AI can improve itself; and in fact chapter 8 is also on this exact subject, though again inconclusive.

Chapter 9, amusingly called “why you still have to fold your laundry,” is the first that explains why some problems are difficult for AI, especially when combined with robotics which is a related but different topic. It is hard, the author states, to get data on “how to physically act in the real world,” especially as we find it challenging to state in miniscule detail how to perform tasks that for us are easy, such as riding a bicycle.

The tenth and final chapter looks at the future. How can we supervise AI when it is in some respects smarter than we are? Again, the chapter is inconclusive, though Rocktäschel says we not be “blindsided by silly mistakes that current AI systems make,” but rather pay attention to the rate of progress.

In the end this book was not quite what I had hoped. It is not really 10 things to know about AI but maybe three or four things, with the discussion spread between chapters in what to me is quite a confusing way. The book also comes over as somewhat defensive about the limitations of AI – but AI does not need defending, it needs explaining.

A good and accessible book on AI.
If you don’t know where to start as an interested party, I highly recommend this book

A solid 3 out of 5 stars. Found it a bit difficult to read at times but in the end it had me asking questions and thinking about what might become of the world and people in 50 years time.

In a growing world of AI, it seems we can’t know enough about the subject to arm ourselves with deeper understanding as it increasingly begins to be introduced into our everyday lives. There are so many benefits to AI but also scary considerations and of course the discussion ‘watif AI takez over the worldddd’ etc..

It’s important to know its applications, restrictions and its uses to our lives and there are many!

This book was great in breaking down information into short and succinct chapters. It’s such an enormous and complicated topic and I can’t claim to have understood it all but it was interesting nonetheless.

I did feel like some prior knowledge may have been helpful here as it was fairly intense in some parts but that’s probably my ignorance more than anything. It’s great at introducing the topic to you and providing you grounds for further reading.

I read it in near one sitting and it covered a broad range of AI. It was really insightful and enlightening in all things related to artificial intelligence.