South Korean Go master Lee Se-Dol recently announced his retirement from professional Go competition. The reason? He felt that no matter how hard he tries, he will never beat AI Go players like AlphaGo. It is a rather sad decision and development of his historical defeat in competition with Google DeepMind’s AlphaGo. It gives the whole thing a more dramatic tone than it should be. However, the defeat of human Go players to AI is neither the end of the world for the Go game nor for the human players. Quite the opposite, I believe this could mean a brand new chapter for the Go game, and an opportunity for us to go back to the original purpose Go game is invented for.
AlphaGo versus Lee Sedol, also known as the Google DeepMind Challenge Match, was a five-game Go match between 18-time world champion Lee Sedol and AlphaGo, a computer Go program developed by Google DeepMind, played in Seoul, South Korea between the 9th and 15th of March 2016. AlphaGo won all but the fourth game; all games were won by resignation. The match has been compared with the historic chess match between Deep Blue and Garry Kasparov in 1997. — Wikipedia
Barring the seemingly dramatic defeat, Lee Se-Dol actually is the only human player actually defeated AlphaGo, even it was only one match. But this defeat deeply hurt the human Go master’s confidence, in an interview with Yonhap News Agency in Seoul, he said
“With the debut of AI in Go games, I’ve realized that I’m not at the top even if I become the number one through frantic efforts,” said Lee. “Even if I become the number one, there is an entity that cannot be defeated,”
To commemorate his retirement, he plans to get into a competition with another AI go player HanDol developed by South Korea’s NHN Entertainment Corp that has already defeated 5 of South Korean’s top Go players.
So what to make of all these? Does this mean Go human players are worthless anymore because AI can already do a much better job? Is Go game not worth playing since a human can never beat a machine? Does this suggest the beginning of the end for human intelligence and the rise for AI to rule them all? The answer is a resounding NO, but to ask why, we need to go back to the original of the Go game.
he Origin Of Go Game
Go is a very ancient game. It’s said that Emperor Yao(尧) of ancient China invented the Go game to teach his naughty son. His son DanZhu(丹朱) was very combative, yet refused to read any book his parents throw his way. How do you talk sense into this kind of kid? Well, entice him to play games of course! It turns out, DanZhu enjoyed the Go game very much. The competitive nature of the game plays into his combative personality, yet to win consistently, he needs to be more patient, smart and strategic. So he gradually learned all these and his parents were much happier.
It’s a rather old and simplified story, but it tells one thing: Go is not created to see who is the best, it is created for educational purposes. I’ve learned to play Go when I was 12 years old and I still remembered our teacher telling us:
“To win the game of Go, you need to be good at calculating the current move, but also thinking long term. You need to accumulate small advantages yet never forget about the big picture. You need to be brave enough to fight the good fight when needed, yet never too reckless if you know you can’t win. When you are behind, you need to have the patience to wait for the best opportunity to strike. When you are ahead, it’s even harder to not get complacent and make mistakes. Doing all these, and you can start to compete in Go game”
The game has so much depth and such a high ‘skill cap’ that no one can claim he has mastered it. It’s said that’s the game played by people in heaven. Myths aside, Go as a game can definitely shape a person’s character, sharpen his mind and strengthen his will. All of these are way more valuable than winning itself.
Go is a Way to Communicate
There is an old saying in Chinese that goes “a person’s Go game style echos his personality(棋如其人)”. It’s easy to hide one’s true nature under well-organized words, but it’s much harder to hide when you’re in a fiercely played game. Go game used to be part of the interview process when ancient Chinese emperors hire his officers. It’s also a way to communicate. Since it’s very popular in ancient China, anyone knows how to read knows how to play Go. Engaging in a healthy competitive Go game is the fastest way to know another person. From your opponent’s playstyle, you’ll quickly know what kind of person he is. Is he conservative and not taking any risks? Is he used to follow some textbook moves or like to think out-of-the-box? After several games, no matter who wins or loses, you’ll know.
Human Beaten by AI on Go, Not a Big Deal
So the million-dollar question is: Does human beaten by AI on Go game change any of the above traits of the game? The answer is obviously no. People can still use Go to cultivate their brain and will. Making friends and finding respect for opponents. It is still a fun game with tons of depth. Winning is not all that important. There is a saying in Go community that goes “A game with a heavy reward can never generate beautiful and brilliant matches(争棋无名局)”. This means if people care too much about winning, their creativity will get surpassed and they will all play safe, resulting in very boring games.
I seldom see people feel bad and say “OMG, I’m not gonna do math anymore because my computer is better than me!”
People just say, “Ok, the computer is better than us on computing numbers. Great. Let’s use it to do the dirty work and let’s focus on inventing new and more powerful algorithms”.
People just say, “Ok, the computer is better than us on computing numbers. Great. Let’s use it to do the dirty work and let’s focus on inventing new and more powerful algorithms”.
The Future of Go with AI will be BrightKe Jie, Photo from Sohu.com
The Lee Se-dol story might be a bit sad, but there is another story related to AlphaGo that is totally in a different tone. Chinese young Go master who was ranked #1 in China professional competition, Ke Jie, also had a three matches game with AlphaGo, and he lost miserably. All three matches were lost and he stood no chance at all. What his reaction? He was obviously in awe and paid respect to his formidable machine opponent, but at the same time, he is intrigued, very intrigued.
He said playing with AlphaGo felt like opening a door to another totally different world. The way AlphaGo played the game is not like anything he encountered before, and he was already a seasoned player and won tons of awards in China and abroad. He then researched thoroughly the three matches he lost, trying to learn from it, and grow out of it. And you know what happened afterward? He had a mind-blowing 12-win streak on his professional competitions, beating every and single human opponent stand on his way.
Different mentality leads to different results. The similar losing games yet different players and different outcomes make us think. Maybe losing to AI is not such a bad thing. We human is always best at learning from past mistakes and getting better.
Adapt and grow, learn and create, that’s something human can definitely do better than the computer, and only the sky is the limit.
The Sky is the Limit
Once we switched our mentality, there are plenty of things Go player can do.
For one, Maybe Mr. Lee Se-dol can join an AI team and try to help develop a new algorithm that can beat AlphaGo. He already planned to play the ‘HanDol’ AI, and maybe in the future, he can start an ‘AI Go Game League’ and get his revenge by beating AlphaGo with a new AI player he helped develop named ‘Se-dol Go’.
Or maybe Mr. Ke Jie can leverage what he learned from AlphaGo and help develop an AI-based Go training system to improve human player’s game. Use AI as a test tool to explore and verify.
Maybe the AI community and Go community should work together to develop different styled AI Go players and have them play in a league, we might be able to see a lot of beautiful games and advance Go game theory.
When AI and Humans work together, only the sky is the limit.