BLACK BOX PARADOX

Chess is a fascinating game that hasn’t lost its appeal even after the strongest human player of the time was finally defeated by a computer. I refer to the historical 1997 face-off between Garry Kasparov and IBM’s Deep Blue. In New York City, Deep Blue won by just a slight amount, 3½–2½. However, this small victory has taught us a lot about chess, computers, and ourselves.

We’ve learned that machines can consistently outperform humans in those competitions (games, conflict, planning activities, etc.) where all the rules are well-defined. No matter how complex these rules and competitions are, given enough time, machines’ superior computational abilities and nearly unlimited memory (storage, really) win against human creativity, intuition, and reasoning.

Since that victory in 1997, chess machines (or chess algorithms) have become stronger with every passing year. Nowadays, even the most brilliant player of our time, Magnus Carlsen, doesn’t have a chance against Stockfish or, even worse, AlfaZero, even if the World’s top ten super-grandmasters aid him.

Further, we also learned that this demonstration of a computer’s ability to imitate intellect in one field doesn’t mean it is getting closer to human-like intelligence. While Deep Blue and all the modern chess programs are brilliant at the one thing they are created for, winning chess games, they are pretty useless for everything else. And to become useful in any new areas, they have to be almost wholly modified.

In fact, this remains a confusing point for those who follow the field of AI by reading articles in popular magazines. These articles create the impression that with every step towards better image or speech recognition, successful Jeopardy performance, or successes in chess or Go, computers and algorithms truly acquire human cognitive abilities.

Modern chess algorithms effectively demonstrate that this is not the case. While being 1,000x (or even 1,000,000x) better than humans in chess and ‘thinking’ 50 or more moves forward during each game, these algorithms are less capable of solving most practical tasks than babies.

Finally, we discovered that despite being unable to compete with computers, we still find the ancient game of chess fascinating, want to play it, and want to succeed at it. Cars are faster than humans, but we still compete in running. Machines are stronger than us, but we still compete in weightlifting. Chess is no different––we will continue to play and compete.

Just a week ago, I finally reached the Lichess.com rating of 2,000. It took me 4.5 years and 11,818 (!) games to do it. Apparently, I have spent 45 days, 9 hours and 38 minutes playing non-stop to get to this point. And I only played when I had time for it.

I wish I had the luxury of playing chess online with anyone worldwide when I was just ten years old. Alas, it wasn’t possible then. To play chess, we needed a chessboard, a clock, and a willing partner, who was often difficult to find (even in Moscow). Some people even played by mail!

The result was that it was hard to play more than a few hundred games per year. Today, with the help of Lichess.com (or Chess.com or a similar platform), I can play on the order of 2,500 games per year without ever leaving my house.

Those who complain about the damage and dangers of technology often overlook simple examples like these. These instances of technology change our lives beyond recognition and make them better and more enjoyable.