You don't need to be a chess expert to appreciate the history of how machines began playing chess. The initial thoughts came from Alan Turing, who proposed the Turing Test (the idea that whether a machine can fool a human into thinking that the machine is human), and Claude Shannon who suggested that the machine can approach chess through either 'brute force' or 'intelligent search'.
Kasparov discusses all technological developments that eventually led to his match with IBM's Deep Blue. The purpose of the book also seems to be a chance for Kasparov to explain his side of the story. I had picked up this book after reading about this match in Nate Silver's The Signal and The Noise. Silver paints Kasparov as a sore loser that couldn't deal with his loss and went berserk claiming that IBM cheated. However, Kasparov seems to give a more fair depiction where he acknowledges his flaws but also uses evidence to outline the unfair lengths that IBM went to for winning.
The book comes across as passionate, just like Kasparov himself. It is incredibly well written! Kasparov discusses how our heuristics (rules of thumb) create biases. For example, in chess one tends to follow "move A must be accompanied by move B" which can limit creativity and progress. Machines and humans each possess certain weaknesses which are complemented by the other. Furthermore, technological development and progress in AI cannot be avoided. So instead of focusing on the doomsday scenarios, we need to learn how to work better with machines. And while using machines for better processes isn't cheating, it can induce a cognitive limp as we heavily rely on digital crutches.
However, the book left me thinking about:
1. How can we be so sure that machines would make us more objective if, as programmers, it is possible that we have programmed our biases into the machine
2. If data was being deleted at every reboot of Deep Blue, then why wait till the end of the match to get the logs If upon being restarted the machine would play a move completely different than what it would have played earlier, then how would the machine logs at the end of the game be of any use This seemed odd considering Kasparov's tactic of finding improvements over his opponents' play.
And what stuck with me even more is this:
"But evolution isn't improvement; it's change. Usually from simple to complex, but the key to it is increasing diversity, a shift in the nature of a thing. Optimization can make a computer code faster but it won't change its nature in creating anything new."