Evolution of Chess Playing Machines

In his book, Mind Master, Viswanathan Anand talks about the evolution of chess playing machines.

“In the beginning the (chess) engines were ridiculously weak… The way they approached a problem was to consider every possible move to counter it, but because they were not looking far enough they would sometimes offer the most ludicrous solutions.”

 

Since computers did brute force calculations of every possible move, they could calculate only X number of moves. Yes, that number X would keep increasing, but that just moved the limit. Such a limit would sometimes led to weird situations:

“For a long time… the machines would go wrong at a point where a long-term decision came into play.”

While computer couldn’t calculate that far out, humans could intuitively feel it.

 

But even with that constraint, over time, humans began to find it increasingly difficult to beat computers. Why?

“Even when we’d outplayed engines for long stretches in a game, we’d end up making a blunder here or a slip there at the final curve, and the machine would beat us… We no longer have the strength to beat them.”

Tiredness, an occasional error. To err is human.

 

As you can see, it was a gradual, slow improvement in how and why computers got better at chess. Which is why this is what Anand says of the time when IBM’s Deep Blue computer beat reigning world champion, Garry Kasparov:

“It was more a milestone in public perception than on the chessboard… Kasparov had beaten it only a year before, so it didn’t make sense for his defeat a year later to be seen as the final verdict on computers taking over.”

Besides, Deep Blue had been finetuned to maximize its chances against exactly one player: Kasparov. It was by no means a generalized program…

 

Then came the era of man + computer v/s another man. In this case, man + machine was clearly superior. For example, in {Anand + computer} v/s Karpov, the result was a crushing 5-1. Here’s how Anand explains the one-sided result:

“It was a wipeout because all I had to do was catch him in a position where he was vulnerable and let the computer take over.”

Once there was a vulnerability, the computer would always find a way through.

 

But with the advent of Artificial Intelligence, things have changed forever:

“(AI’s) beauty lies in the idea of self-learning from examples without prior knowledge, which eliminates human bias and intervention. It took AI’s program AlphaZero just four hours from being led to with nothing but the rules of the game for it to destroy the highest-rated chess engine Stockfish in a 1,000 game match.”

It is with AI programs that computers have truly vanquished humans.