American Go E-Journal » Computer Go/AI

Chinese Professionals React to the Historic AlphaGo Win

Friday January 29, 2016

 The following is translated by Jennie Shen 2P from a WeiqiTV video2016.01.29_Fan-Hui

Liuxing 7p: before the games, what were you expecting?
FanHui 2p: I was thinking about how to torture/destroy the program, and see how many handicap stones I could give to it.
Liuxing 2p: if you play AlphaGo again, do you have confidence?”
FanHui 2p: I can’t tell you this, it’s a secret. but I can tell you how I felt when I was beaten up by AlphaGo. The first game, I wanted to play a peaceful game, but AlphaGo played an endgame tesuji, I found out I didn’t have a chance. The second game, I was thinking, maybe AlphaGo is just good at endgame, so I should fight. I got a good result after the avalanche joseki, but screwed up at the lower right corner… So, later, I realized, well, humans have emotions, the emotions will effect the game a lot, but this guy (AlphaGo) doesn’t have emotions, nothing affects it, it won’t make mistakes. The 3rd and 4th game, I was already completely destroyed mentally, no confidence at all. Lots of people asked me these questions: were you in bad shape? is this your full strength? some people even asked if I was sandbagging? here’s what I want to say: for humans playing Go, lots of things matter, like emotions, mentality, we can’t play our full strength. I tried my best…but it’s really really painful every time when I lost to Alphago. I felt like the world was turned upside down, it’s ridiculous.”
Shiyue 9p: When I saw the news yesterday, I couldn’t believe it was true. But the 5 games impressed me a lot, I think AlphaGo has pro level, but still there’s a distance from top pros. I don’t know how does machine study by itself, but I know it studies and improves, I think it won’t be easy for Lee Sedol in March. To us pros, we can learn from the program. If the program gets to top level, I will try different moves to play it, programs are tools, depends on how we use these tools. Some people think Go is culture, it’s mysterious, they don’t want to see AI beats human, some people think we can see a bigger picture and the depth of Go with program’s help. It’s our choice.
If AI beats humans, what does this mean to us?
LiZhe 6p: if AI beats human, maybe in March, maybe within a year, it will change the human Go world. We need to face these big changes. For example, as the competition Go weakens, people might focus more on the culture of Go. In the future, AI will help us to learn Go skills, people will start to think about what Go brings to us? Go will become a way of communication.
Kejie 9p : The 5 games were not that impressive, but the scary part is, AI improves. I use to think AI can never beat human, at least it won’t happen within 10 years. but this is unbelievable. When I was looking at the games, I didn’t know which is human, which is machine, I can’t tell the difference, AlphaGo has very good sense of balance. I think Lee Sedol will win the match in March. I really want to play Alphago, not just me, all the active Chinese pros want to play it. I have confidence to win now, but it’s hard to say later. I will still play Go even if AI beats human in the future.
Check out Myungwan Kim 9Ps January 28 commentary on the AlphaGo-Fan Hui games on the AGA YouTube channel.

Share
Categories: Computer Go/AI,World
Share

Myungwan Kim 9P to Analyze Fan Hui-AlphaGo games this Friday

Thursday January 28, 2016

Myungwan Kim 9P will analyze the Fan Hui-AlphaGo games on the AGA YouTube and TwitchTV channels this Friday at 9p EST (6p PST). And for the first time, the broadcast will be simulcast on weiqitv.com in China. Yesterday’s news2016.01.28_nature-cover that Google’s go-playing AI, AlphaGo, had swept Fan Hui 2P 5-0 rocketed around the world, receiving international coverage including write-ups in the New York Times, Bloomberg News, the BBC, Wall 2016.01.28_myungwan-kimStreet Journal, MIT Technology Review, Wired, NPR and blowing up on the AGA’s  Facebook and Twitter feeds.

“I was shocked at how Alphago played,” Kim (left) told the E-Journal. “It played like a human professional. I am sad that this computer program might beat me, but I don’t think it can beat Lee Sedol. I will tell you why in my commentary.” Google DeepMind, the British artificial intelligence company which developed AlphaGo, has issued a challenge to Lee Sedol 9P from South Korea, the top player in the world for much of the last 10 years, to play a 5-game, million-dollar match in March. Andrew Jackson will host the livestream broadcast.

Share
Categories: Computer Go/AI,World
Share

Google and Facebook Race to Solve Go

Sunday December 13, 2015

“Inside companies like Google and Facebook, deep learning is proving remarkably adept at recognizing images and grasping spacial patterns—a skill well suited to Go,” reported Cade Metz in Wired last week . “As they explore so many other opportunities this technology presents, Google and Facebook are also racing to see whether it can finally crack the ancient game. As Facebook AI researcher Yuandong Tian explains, Go is a classic AI problem—a problem that’s immensely attractive because it’s immensely difficult. The company believes that solving Go will not only help refine the AI that drives its popular social network, but also prove the value of artificial intelligence. Rob Fergus, another Facebook researcher, agrees. “The goal is advancing AI,” he says. But he also acknowledges that the company is driven, at least in a small way, by a friendly rivalry with Google. There’s pride to be found in solving the game of Go.” For more on Facebook’s research, check out “How Facebook’s AI Researchers Built a Game-Changing Go Engine” in the MIT Technology Review last week.

Smart Go’s Anders Kierulf also recently published a blog post on “Go at Facebook”, saying that “As long as Facebook and Google stick with trying to find general solutions to general problems, I don’t think top Go programs like Zen and Crazy Stone have anything to worry about. But once these giants decide to beat the strongest human players and are willing to focus on Go-specific solutions, it will get interesting.” Check out also his explainer on Monte Carlo Tree Search and his post about the impact of the Swift programming language on his go programs.

 

Share
Categories: Computer Go/AI
Share