A highly anticipated Go match will finally begin between Korea’s Go genius Lee Se-dol and the artificial intelligence system “AlphaGo” on March 9. Lee is expected to romp with a winning percentage of over 72 percent. What is behind Google’s challenge to win a human being in a highly publicized Go game called one of the essential parts in 5,000-year-old spiritual culture of the East?
Google announced in a meeting with the press at Four Seasons Hotel on March 8, the eve of the first match day that Lee will play with AlphaGo for five times from March 9 to March 15. AlphaGo is a Go game program of DeepMind, the Artificial Intelligence (AI) arm which Google took over in 2014.
“Regardless of the results of the match, the winner will be humankind,” said Eric Schmidt, chairman of Alphabet (the holding company of Google) who surprised people in the meeting by appearing without a notice. “The development of artificial intelligence and mechanical learning will make each and every human being smarter and more competent.
Chairman Schmidt expressed his own predictions about the game. “AlphaGo will not easily succumb. I expect a fifty and fifty chance of AlphaGo’s winning,” he said with a strong confidence in AlphaGo’s technology.
But Go pro Lee showed an unassuming attitude, saying, “I have asserted that I will win all of the five matches. But today, after I got some understanding of the algorithm of AlphaGo’s artificial intelligence, a humane mistake by me may force me to lose one of the five matches to the AI system.”
Excluding Chairman Schmidt, many experts expected Lee to win the event with a likelihood of 70 percent to 100 percent. US$1 million is at stake in this best-of-five human VS machine Go event. In addition to the prize money, Lee will be receive US$150,000 for the event and US$20,000 per win. If Lee win all of the matches, he will be able to take home about 1.5 billion won.
Many experts, however, say that the final winner will be neither Lee nor AlphaGo but Google even though AlphaGo loses. This is because the internet search giant will gain astronomical profits by broadcasting the event via YouTube and publicizing its artificial intelligence software named “AlphaGo” to people around the world.
On top of that, Google will be able to save more data to boost the intelligence of AlphaGo by learning from the match with the world-class pro Go player. Go is the final stage of brain games between human beings and artificial intelligence. Except for Go, the world’s top players in chess, quiz and Oriental chess already have surrendered to artificial intelligence programs.
“This match will make AlphaGo smarter since it will obtain the data of its match with the world’s top Go player,” said Kim Dae-shik, a professor of the KAIST. “If the two play a second match, AlphaGo will be much stronger.”
Google’s ultimate goal of artificial intelligence development is to apply AlphaGo technology to the health and medical sector, going beyond games. “Although games are fun and I am interested in Go matches between AlphaGo and human beings, we want to apply this technology to more than games,” said Demis Hassabis, CEO of DeepMind. “AlphaGo has many applicable areas such as health care, robots and smart systems. But the most interesting sector is heath care and medical services.”