Google and Korean Baduk (Chinese Go) Association (KBA) today held the “Google Deepmind Challenge Match Press Briefing” to announce the details on the upcoming games between Sedol Lee (9-dan) and AlphaGo, a computer AI by Google DeepMind.
Vice Secretary of KBA, Cheemun Park, and Sedol Lee attended the briefing, and Demis Hassabis, the CEO of Google DeepMind joined on live video to announce the match location, time, rules, and broadcasting details.
The series of matches will consist of a total of five rounds, the winner of which takes $1 million USD. Should AlphaGo win, the entire sum will be donated to UNICEF in their programs supporting STEM education.
The matches will be run according to Chinese rules of play, with each player getting two hours to carry out their moves. Once each player has used up his two-hour limit, he will be on 60-second countdowns to execute each move, with just three chances to exceed that limit to deliberate. The entire match is expected to last 4~5 hours.
Every round of the Google DeepMind challenge will be broadcast live through Google DeepMind’s channel on YouTube. In Korea, China, and Japan, the matches will also be available on live TV.
All matches will have official commentary in both English and Korean. The English commentary will be done by Michael Redmond, the only western Go player to have reached professional 9-dan (the highest level). As for the Korean commentary, Chang-hyeok Yu, the current coach of the Korean national team will be in charge, followed by Seongryong Kim, Taegon Kim, and Hyeonwook Lee.
David Hassabis, CEO of Google DeepMind, remarked that “Go is the most sublime game that humans have invented. The beautiful complexity of Go derives from the elegant, simple rules of the game. Since intuition and visceral feelings often triumph over brute calculation of the possible cases, Go is a difficult game for computers to master. We are honored that DeepMind was able to invite Sedol Lee, a legend, and are looking forward to the upcoming matches. Regardless of AlphaGo’s results against Lee, we hope to spark interest around the world for the game of Go,”
Lee commented, “these matches are the first time a computer will be facing off a human without handicap, and I am happy to be participating in such an important match. No matter the outcome, the face-off will be a milestone in the history of Go. I’ve heard that the DeepMind’s AI has had incredible progress, but I’m still confident of a victory.”
The number of outcomes of a Go game exceeds the number of atoms in the universe, rendering brute-force counting of the possibilities useless. For this reason, the game has been consistently used to test the cutting edge of computer AI for decades. Google DeepMind introduced the latest version of AlphaGo just last month on the science journal Nature, as the first AI expected to win against a human without handicap.
All five matches in the Google DeepMind challenge will be held at 1PM throughout 3/9~3/15 at a special venue in the Four Seasons Hotel Seoul.
Match 1: 3/9 (Wed)
Match 2: 3/10 (Thurs)
Match 3: 3/12 (Sat)
Match 4: 3/13 (Sun)
Match 5: 3/15 (Tues)
– Inspired by Platum