The 2016 Go match between Lee Sedol 9 and AlphaGo was a battle of human and AI pride, and the reason people cheered Lee Sedol’s victory was probably due to the relief of confirming the superiority of human intellect. To overcome this unconscious anxiety about the development of A.I., we need to find a role for humans that A.I. cannot play.

Lee Sedol vs AlphaGo

A long time ago, the five-round Go match between Lee Sedol 9 and AlphaGo, which took place from March 9 to March 15, 2016, captured the attention of people around the world. It was not just a game of Go, but a battle of pride between a human player and an artificial intelligence computer (sometimes referred to simply as a “robot”). Lee Sedol’s successive defeats in the first three countries made us sigh and say, “Humans can’t beat computers anymore,” while others praised the perfection of A.I. robots and their technological prowess.
However, after the victory of Lee Sedol 9 in four countries, all kinds of media such as the Internet, TV, and social media cheered that ‘humans have surpassed computers’. This was many times more enthusiastic than during AlphaGo’s winning streak, and I wondered if it was not just the joy of victory, but also the relief of defending the title of “top tier of intellectual level”. In this article, I would like to discuss the irony that while we strive to create an AI that can think like a human being, we are more likely to celebrate a human victory than an AI victory.

The rise of robots and AI

The term artificial intelligence (AI) was first used in 1956 at a meeting of ten scientists at Dartmouth College in the United States. AI research, which began in the 1950s at universities in the United States, was initially focused on enabling machines to take over not only physical labor but also human intellectual activities. The research progressed by leaps and bounds in the 1970s, with additional techniques and algorithms being developed, and by the 1980s, expert systems that could match the intelligence and experience of human experts in certain fields were being realized, but AI was still primarily intended to take over complex calculations and tasks that humans could not perform. It might have been thought of as a human assistant.
It wasn’t until the early 1970s when humanoids began to emerge that this “assistant” role became “human-like”. With the advent of the first humanoid robot, Wabot1, developed in Japan in 1973, the end goal of robots became to behave like humans, and AI began to develop with the goal of thinking on a ‘human-like’ or ‘human-like’ level, rather than just doing chores for us. At the time, the results of these clumsy attempts to imitate humans may have seemed curious and entertaining, but no one could have imagined that the programs that performed arithmetic or did the work for them would evolve to the point where they could think as well as, or even better than, humans like AlphaGo.

Human vs. AI

The 1990s saw another boom in AI research with the development of the Internet. Despite the difficulty of collecting and supplying a large amount of knowledge and information, the Internet’s availability of search engines made it possible for AI to learn on its own, a form of machine learning. It was at this point that AI that could learn on its own without human assistance was thought to have reached a level close to human intelligence, and the battle between humans and AI began.
In 1996, IBM’s supercomputer Deep Blue played world chess champion Gary Kasparov in a match that Kasparov won, but in 1997, an upgraded Deep Blue defeated Kasparov by two wins, three draws, and one loss. Later, IBM’s supercomputer Watson appeared on the popular American quiz show Jeopardy and defeated human champion Ken Jennings, demonstrating that artificial intelligence can outperform humans in areas other than chess.
While AI computers have a significant advantage in these games because they can win by knowing the number of possible moves, it has been predicted that they will not be able to outperform humans in Go. The number of possible moves in chess is 120 to the power of 10, but the number of possible moves in Go is much larger, more than 170 to the power of 10, due to the complexity of the rules, such as judging formations and removing and replacing dead stones. However, modern technology has created a monster called AlphaGo, and it gave us a taste of what artificial intelligence has reached by defeating the world’s top player, Lee Sedol 9. We were awestruck by AlphaGo’s abilities, disappointed by Lee Sedol’s defeat, and enthusiastically cheered for a valuable win. But was this ‘cheering’ solely out of the joy of victory? I suspect that part of the cheer came from a sense of relief that humans have demonstrated that there is a range of intellectual capabilities that AI cannot yet surpass.

The future of AI and the role of humans

In 1950, American author Isaac Asimov proposed what he called the “Three Laws of Robotics” in his novel. To summarize them briefly, robots should not be able to harm humans under any circumstances. While this is part of fiction, it seems to recognize the possibility that there will be situations where robots will outgrow humans and cause harm. The robot HAL9000 in A Space Odyssey (2001), a movie that shows a form of artificial intelligence that refuses to be controlled by humans and acts autonomously by processing commands at will, and I, Robot (2004), a movie that depicts a situation where an artificial intelligence robot that acts out of control and threatens humanity is quite realistic.
It may be a bit of a leap to compare this to simply chess or Go, but who would have thought that we could have gone to the moon? The rise of popular culture with such ideas has made us aware of the possibility of such a situation coming without us even realizing it, which is why we cheered even more enthusiastically for Lee Sedol 9, who defended his title at the top of the intellectual heap.
In order to resolve the irony of aiming for AI that is close to humans, but also to resolve the unconscious anxiety of AI that is beyond humans, and to ensure that we are not truly usurped in the distant or near future, it is important that along with the development of artificial intelligence, we also find out what humans cannot do, no matter how advanced AI is.