Go Grandmaster Lee Sedol and Honorary 9-Dan Go Pro “AlphaGo”
Google's AlphaGo sealed 4-1 victory over world Go champion Lee Sedol. When AlphaGo, which had already bested European Go champion Fan Hui in five games out of five, won the first three games of the historic five-game match against Lee Sedol, the majority view was that human Go masters are no match for artificial intelligence (AI). Lee Sedol, however, struck back with a vengeance to score his first win against the supercomputer in the fourth game, which was an unexpected turn of events.
Yet it seems obvious that it’s getting hard to outsmart computers. Nevertheless, it is also difficult to make computers completely flawless. As a matter of fact, software is written by humans and thus achieving 100% mastery of all subjects is impossible
The first man vs. machine competition in Jeopardy! history showed just how difficult learning was for AI. Facing off against Ken Jennings and Brad Rutter, IBM's Watson computer answered 66 correct and 9 incorrect. Interestingly, where Watson failed was more instructive than when it succeeded.
Watson was the only contestant to miss the Final Jeopardy! response in the category US Cities. In response to “Its largest airport is named for a World War II hero; its second largest, for a World War II battle” in the category of “U.S. Cities,” Watson said “What is Toronto?”
Watson also tripped up on an "Olympic Oddities" question. It was about George Eyser, a German-American gymnast who competed in the 1904 Summer Olympics, earning six medals in one day, three gold, two silver and one bronze medals. In response to “It was the anatomical oddity of U.S. gymnast George Eyser, who won a gold medal on the parallel bars in 1904,” Ken Jennings, who won 74 Jeopardy! games in a row, said “Missing a hand (wrong).” Watson answered, "What is leg?" (wrong). The correct answer was “Missing a leg.” Watson was simply fed mountains of information. Watson wasn’t able to follow the context created by Jenning's answer. Therefore, its answer of "What is a leg?" rather than "What is missing a leg?" was actually deemed incorrect. It clearly showed that there was a limit to AI.
Computers and humans have their own strengths in different areas. Thus, Moravec's paradox - the notion that the hard problems are easy and the easy problems are hard- was discovered. The mental abilities of a four-year-old that we take for granted – for example, describing what's going on in a photo – are the hardest for computers to acquire. In contrast, the world's most complex board game isn’t so difficult for computers.
Though great progress has been made in AI, teachers like Lee Sedol are needed to take it to new heights. However, Google is unlikely to arrange more AlphaGo vs. human matches. After Deep Blue defeated world chess champion Garry Kasparov and Watson beat Ken Jennings, IBM put an end to its AI publicity stunts. Through the two highly publicized human vs. machine matches, IBM succeeded in drawing more than enough publicity to AI.
Google chose South Korean Go grandmaster Lee Sedol because he is one of the world’s top GO players who won 18 international championships. The level of publicity surrounding AlphaGo’s 4-1 victory over Lee Sedol far exceeded Google’s expectations. Drawing on such momentum, Google may have wanted to tout safe self-driving car technology. Showing that its AI technology is sophisticated and crafty enough to outwit a top-ranked Go professional, Google successfully caught the eyes of consumers and cleverly gave its stock price a boost.
If AlphaGo hadn’t lost the fourth game, Google would be much happier now. The unexpected defeat was humiliating for Google. Cornered by Lee Sedol, AlphaGo made a series of bad moves to the disappointment of AI supporters. AlphaGo should have resigned before exposing such fumbling moves to the world. AlphaGo’s earlier resignation would have drawn more cheers and applause.
The Lee Sedol vs. AlphaGo match will go down in history. Lee Sedol’s valiant fight against 1,202 CPUs came as a breath of fresh air in this spring and has proven that humans are still worthy opponents to AI. We’re just grateful for that.