Google AI AlphaGo vs Lee Sedol (2016-03)
This is a historical moment. Human vs Machine, and Machine won, 4 to 1.
About the Match
- AlphaGo versus Lee Sedol.
- 5 games match. March 2016. At Seoul, Korea.
- Each side has 2 hours, after that, 60-second per move. Cannot exceed 3 times. (called byo-yomi period)
- AlphaGo is a computer program from Google, running on remote machines of 1920 CPUs and 280 GPUs thru network.
- Lee Sedol is multiple-times world champion of go. (by one rating, Lee Sedol is currenty word's 4th.)
- The winner takes $1 million prize. If AlphaGo wins, the prize will be donated to charities, including UNICEF. Besides the monetary prize, Lee Sedol will receive at least $150,000 for participating in all the five games and additional $20,000 for each win.
About AlphaGo
AlphaGo is a computer program developed by Google to play the board game Go.
AlphaGo's algorithm uses a combination of neural network and tree search.
[see Artificial Neural Network is Nasty]
Try a Game of Go
if you never tried go before and is curious, yes, try it. It is necessary to play a few games to get a taste of it, even on a small 9x9 board.
you can try the gnu go software for this purpose. see [see Great Math Board Game Software]
but, be warned, it's a game that can suck years of your life. (i assume you are programer or college student of math or such.) It has, indeed, sucked about a year of my youth. Whether the time is wasted is a point of view.
Game Summary
5 games. Lee won 4th game, lost all others.
Go Player Ranking
As of 2016-03-11, top player is Ke Jie (Chinese name 柯洁. Born 1997), according to the site Go Player Elo Ratings http://www.goratings.org/
- 1, Ke Jie 柯洁, ♂, cn, 3621
- 2, Park Jungwhan 박정환 朴廷桓, ♂, kr, 3569
- 3, Iyama Yuta , 井山裕太, ♂, jp, 3546
- 4, Lee Sedol 이세돌, 李世乭, ♂, kr, 3521
- 5, Shi Yue 时越, ♂, cn, 3510
- 6, Kim Jiseok 김지석, 金志錫, ♂, kr, 3504
- 7, Park Yeonghun 朴永训, ♂, kr, 3503
- 8, Mi Yuting 芈昱廷, ♂, cn, 3502
- 9, Zhou Ruiyang 周睿羊, ♂, cn, 3498
- 10, Kang Dongyun 姜东润, ♂, kr, 3498
Video Commentary of Games
the host of the commentary, guy on the left, is Chris Garlock. Really annoying guy, with his trivial, humbleness-showoff, no-content, slow drone out blatherings.
the guy on the right is Mike Redmond, the only westerner to attain 9dan. He seems cocky and insecure. He's also somewhat annoying.
The official English commentary is lame and boring to watch, and long. Remember, each game is about 4 hours.
Go Ranking System
Alternative English Commentary
here's commentary by Myungwan Kim. Myungwan Kim (金明完) is a 9 dan pro, from Korea. http://senseis.xmp.net/?KimMyungwan
Chinese Commentary by Ke Jie 柯洁解说
Ke Jie is currently top rated player in the world. He's only 19.
The following are much fun to watch, but they are in Chinese. (it's a TV show. Also, in Asia, it's common to have females who are professional go players.)
Go Game Record, Format, Software
the commonly used file format to store go games is SGF.
It is plain text file format.
list of sgf file readers
- http://gobase.org/software/editors/
- http://senseis.xmp.net/?SmartGameFormat
- web based JavaScript viewers http://senseis.xmp.net/?WebViewers
Game Record Files
- AlphaGo_vs_Lee_Sedol_game_1_2016-03-09.sgf
- AlphaGo_vs_Lee_Sedol_game_2_2016-03-10.sgf
- AlphaGo_vs_Lee_Sedol_game_3_2016-03-12.sgf
- AlphaGo_vs_Lee_Sedol_game_4_2016-03-13.sgf
- AlphaGo_vs_Lee_Sedol_game_5_2016-03-15.sgf
Impacts on Humanity
in 1997, when machine beats human in chess, that was a big event. Even a bigger even than today's AlphaGo.
now it's 20 years.
- Did human go to mars?
- Do we still have homeless?
- Do we still have war?
- Are politicians now honest?
there isn't any drastic improvement of humanity. Actually, there isn't much of a firm, solid, confirmation by some scientific measures, that human animals are doing better. (i think crime rate, have decrease slightly. But i don't think much improved in bigger scale of things.)
this Google DeepMind thing, is exciting to tech geeks as how pop song is exciting to pop idol followers.
on the positive side, I think this event can be a representative milestone of today's Artificial Intelligence progress, many of which we've been using for a couple years now. Face recognization, Apple Siri, google search, self-driving cars, robotics.