Games computers play3 min read . Updated: 24 May 2016, 07:50 PM IST
The author on how an artificial intelligence (AI) program beat a human at an ancient Chinese board game
In March, the world’s finest Go player accepted the challenge to play against a computer program. The human player was Lee Sedol of South Korea, a grandmaster, and challenging him was an artificial intelligence (AI) program called AlphaGo, developed by Google DeepMind, a British AI company. They were to play five games, and the prize money was $1 million (around ₹ 6.7 crore).
Go is an ancient Chinese game that is played on a square board slightly larger than a chessboard. Two players use flat, round pieces called “stones" that look like large buttons. One player uses black stones, the other white. Like chess, it’s a game of strategy: The idea is to capture maximum territory on a board that usually has a 19x19 grid.
It is not an easy game to play; in fact, it’s vastly more complex than chess. The co-founder of DeepMind, Demis Hassabis, has been quoted as saying that “there are more possible Go positions than atoms in the universe".
Out of sheer curiosity, I decided to watch the first game (all the games were streamed live and are available on YouTube). In any game, you try and guess how the opponent’s mind works, you try and anticipate their moves. I wondered how a machine would fathom the mind of a human player who thinks several moves ahead. My bets were on Lee.
This is how the game I watched was played: On one side of the Go board sat Lee. Facing him, across the table, was one of AlphaGo’s programmers, Aja Huang, who had a computer monitor logged in to the AlphaGo server. As Lee played his move on the board, Huang would use the mouse and repeat the move on the monitor. After a pause, AlphaGo would reveal its move on the monitor, and Huang would carry it out on the board.
It was a classic human vs AI contest, and watching it was a mind-blowing experience. On a split-screen, a professional American Go player gave a running commentary, and as the game began, he said, “History is being made here."
AlphaGo won 4 to 1. “I misjudged the capabilities of AlphaGo and felt powerless," Lee said after the game.
I don’t know enough about Go; I made an attempt to learn the game several years ago, but soon gave up. However, I got turned on by this incredible match, and downloaded several apps to try my hand at the game once again. I found a lot more helpful and interactive stuff online than I had six years ago.
Both the American Go Association (www.usgo.org) and the British Go Association (www.britgo.org) have tutorials for beginners. There are several other websites where you can learn the rules and play some basic moves against the computer; you will have to find your comfort level.
I use a Mac and have an app called Goban that I downloaded from the iTunes Store (there are other links from where you can download). I used to have it when I first decided to learn the game. But if you have an iPad, there’s a fine comprehensive app called Go Books that you can download for free, and then buy an e-book from the several listed on the app—depending on the level of your game.
The ones recommended to me by a Go enthusiast are Learn To Play Go by Janice Kim ( ₹ 420), and Graded Go Problems For Beginners ( ₹ 420). And once you know the basic rules, try the Tsumego Pro app (again, recommended to me) that has a large collection of Go problems that you can attempt on an iPhone, iPad or Android device.
In my renewed enthusiasm, I find myself using the Tsumego app—there’s a red light when I make a wrong move, and I can also get a hint. But the more I try my hand at the game, the more I think that the only way I will improve is to find a human partner who will tell me why I am going wrong. Online games are fine for those who know the basics pretty well.
Which brings me back to AlphaGo. The outcome of the match has been called “a milestone in the history of artificial intelligence", and it does make you wonder: what next? There’s an interesting line in the sci-fi movie Ex Machina, in which a scientist builds a beautiful female humanoid with artificial intelligence. He says: “One day the AIs are going to look back on us the same way we look at fossil skeletons on the plains of Africa."
Shekhar Bhatia is a science buff and a geek at heart.