The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World,
I read this book on the recommendation of a friend. We are both computer scientists, with her being vastly more knowledgeable than I am on the topic of machine learning. In fact, I barely know machine learning at all. With that as the background, reading this book is a good introduction to what’s been happening out in the world - provided you already had an idea about it. If you’re a beginner or someone who doesn’t already know about machine learning, I think the book is going to be unbearably dense and confusing.
As for the contents of the book itself, I found the first half exceedingly enjoyable and fun to read. The little titbits of history, of explanations, and the connections were all an excellent addition to the reading. However, the second half of the book, in particular the last two or three chapters pertaining to the master algorithm itself were woefully inadequate, short, and dense for my continued comfort.
I would still recommend reading this book for the topic and perspective, but with a fair warning about taking the last chapters on application of machine learning as ground truth. They’re not (hopefully). For instance, the author thinks privacy is overrated, laws that limit misuse of data are myopic, and that sooner or later we will all be handing our data to a ‘trusted third party’. I cannot agree with this in good faith. I must point out the good along with the bad - there was a good argument for transparency in the workings of algorithms, about introducing control points for tweaking, and being more cohesive in the different branches of technology and its applications (finance, health, medicine, even dating). These I agree and support, but lest we do not forget that accountability and ethics are also what makes humans as humans in addition to our superior intellect and reasoning capabilities. ,