“In the era of AI, we need people who can create Go, not just those who can play it well. To create Go well, we must collaborate with artificial intelligence (AI). While it's not easy for humans to be free from fixed ideas, AI can play that role.”

Go player Lee Se-dol, ranked 9th, has become a distinguished professor in the Department of Mechanical Engineering at Ulsan National Institute of Science and Technology (UNIST). Starting this semester, Professor Lee will teach a course on 'Board Game Creation for Scientists,' holding classes for 6 hours every other Friday. Park Jong-rae, president of UNIST, noted, “Professor Lee has sparked discussions about the relationship between AI and humans, and his insights provide significant help in cultivating talent with creative insight, as he provided a move that was impossible for AI.”

Ise-dol, a special professor at Ulsan National Institute of Science and Technology (UNIST), said at a press conference on Nov. 11, "We need someone who can create board games like Go rather than someone who plays Go well." /Courtesy of UNIST

In March 2016, Professor Lee lost four times and won once against the Go AI 'AlphaGo' created by DeepMind in their historic match. The 78th move he made in the fourth game, in which he defeated AlphaGo, has been regarded as 'the divine move.' Following this brilliant move, AlphaGo made a series of mistakes, and Professor Lee calmly responded until the endgame, securing a precious victory. President Park viewed Professor Lee's move that beat AlphaGo as representative of the knowledge and inspiration we need in the era of AI.

During a press conference held on the UNIST campus in the engineering building on the 11th, Professor Lee reflected on his matches against AlphaGo. He said, “After the first and second games, I realized it would be hard to win normally, so I devised a strategy from the third game onward. I approached the match with a strategy to avoid the strong early and late game, aiming to determine the outcome between moves 50 and 100.”

He revealed that the 68th move prior to the 78th was the decisive one. Professor Lee stated, “The 68th move would have been a bizarre move that should never be played against a human opponent, but it induced AlphaGo’s bug (error) from there. The decisive move worked because AlphaGo made a mistake at the 78th move.”

Scene 1 = The white stones scattered in the center are blocked from retreating. If these white stones are captured, it becomes difficult for white to win. At a time when a miraculous move for a comeback is necessary, the move of white 1 was crucial. This move was literally a miraculous move for a comeback. In the situation where black 2 through 5 proceeded, the taste of pushing with A and the taste of cutting and bringing the center to life with B allowed white's escape to succeed and became the top contributor to victory. Scene 2 = Now let's take a look at AlphaGo's 'Tteoksoo Series.' The moves from AlphaGo's 1 (total move 93) to 9 (total move 101) were all strange moves. In particular, move 5 was a problematic move that would make one doubt their eyes. It was even stranger as it was not a situation for extending time. Scene 3 = AlphaGo's strange moves continued until the end. The move made with 1 (total move 161) and the move made at 3 (total move 167) that provoked 4 were all strange placements.

Professor Lee observed that with the advancement of AI after AlphaGo, such error-inducing approaches are no longer valid. However, he explained that they gained clues about the unique creativity that is hard for AI to imitate through their matches against AlphaGo. Professor Lee remarked, “After the match with AlphaGo, I began to question the term ‘creative’ and wondered if the AI, having learned from human Go, might actually make moves that seem more creative than a human.”

He found answers outside of fixed ideas. Professor Lee explained, “I thought I was trying not to get trapped in a framework while playing Go, but I found myself caught in a framework without realizing it. However, since AI is not trapped in such fixed ideas, it appeared to be creative.” He added, “By observing moves from AI that break free from fixed ideas, humans can also learn to escape from their fixed notions and frameworks.”

After retiring as a Go player, Professor Lee also ventured into board game creation. A board game involves physical tools such as boards made of paper, wood, or plastic, as well as cards, dice, and pieces. Go and chess are examples. Professor Lee has developed and released three types of board games. He stated, “In the future, we will not need people who can play Go well, but those who can create it, and we must strive to create Go well in collaboration with AI.”

On this day, a board game creation class taught by Professor Lee was also conducted in Engineering Building 5 at UNIST. The class consisted of 24 students working in groups of four to create actual board games. Professor Lee remarked, “There is no better way to feel intuition, insight, and creativity than making a board game, and I intend to provide insights on how creative thinking of AI and humans can harmoniously coexist.”

Ise-dol is conducting a lecture on creating board games for scientists. /Courtesy of UNIST

President Park also unveiled plans for the UNIST AI smart campus. The invitation of Professor Lee was part of the initiatives for the AI smart campus promoted by President Park. The AI smart campus will integrate advanced AI technology across education, research, and administration, and provide practical AI training focused on the industrial sector in the Ulsan area. UNIST plans to provide AI application education for faculty and students and establish infrastructure including a small language model (sLM) and an autonomous laboratory.

President Park stated, “This year marks the inaugural year for building the UNIST AI campus, where we will implement ‘AI basic mandatory education’ for all students, regardless of their majors, to ensure they acquire fundamental knowledge of AI, while also establishing a ‘one person, one AI creation’ framework to innovate the entire process of learning, research, and creation.” He added, “We aim to play a pivotal role in transforming the industrial structure centered on manufacturing in the southeastern region into a high-value-added industry based on AI.”