Tak Mampu Saingi Kecerdasan Buatan, Pemain Go Profesional Memilih Pensiun

Sejak diperkenalkan ke publik pada 2016, AlphaGo—kecerdasan buatan yang memainkan Go Google—berulang kali memenangkan kompetisi. Lee Se-dol adalah satu-satunya manusia yang pernah mengalahkannya. Namun, sang grandmaster Go kini memutuskan pensiun karena sadar program AI terbarunya takkan mampu ditundukkan.

"Sejak game Go menggunakan AI, saya menyadari takkan mampu menduduki peringkat teratas bahkan jika saya menjadi pemain nomor satu karena bermain dengan panik," kata Lee kepada Yonhap News di Korea. "Entitas ini tidak bisa dikalahkan."

Lee mengalahkan AlphaGo dalam pertandingan pertamanya pada 2016, tetapi kemudian empat kali dibantai AI. Sejak itu, Google melatih versi terbaru AlphaGo dengan melawan versi lamanya.

Versi AlphaGo baru bisa mengalahkan pendahulunya setelah memenangkan 100 game berturut-turut dari 4,9 juta game. Tim pengembang menggunakan teknik serupa untuk menciptakan AI yang mampu memainkan Starcraft II lebih baik daripada 99,98 persen manusia.

Sadar dengan kekuatan yang dihadapi, Lee pun hengkang dari Go. Namun, Lee akan bertanding untuk terakhir kalinya melawan HanDol—AI serupa yang dikembangkan NHN Entertainment di Korea Selatan—sebelum undur diri. HanDol telah menaklukkan lima jagoan Go di Korea Selatan.

Supaya adil, Lee akan memulai pertandingannya dengan dua batu yang sudah ada di papan untuk melumpuhkan HanDol.

"Saya yakin bakalan kalah bahkan dengan keunggulan two-stone sekalipun," tutur Lee. "Saya enggak pernah mengikuti berita Go lagi. Saya kan sudah pensiun, jadi ingin bermain santai dengan HanDol."

Since its debut in 2016, AlphaGo—Google’s Go playing AI—has repeatedly crushed the competition. Only one person has ever defeated the AI, Go master Lee Se-dol. Now, faced with an improved version of the AI program, Lee says he’s retiring.

"With the debut of AI in Go games, I've realized that I'm not at the top even if I become the number one through frantic efforts," Lee told Yonhap News in Korea. "Even if I become the number one, there is an entity that cannot be defeated.”

Lee beat AlphaGo once during their first matchup in 2016, but lost four other games. After the games, Google trained a new version of AlphaGo by forcing it to play against its predecessor.

The old and young AI fought over the course of 4.9 million games until the new AlphaGo was able to beat its predecessor 100 games in a row. The same team used the same technique to create AIs that play Starcraft II better than 99.98% of human players.

Faced with such impossible competition, Lee is leaving Go behind. But, to commemorate his retirement, he’s playing a final set of games against HanDol—a similar AI created by South Korea’s NHN Entertainment. HanDol has already bested five other South Korean Go champions.

To keep things fair, Lee will start his games against HanDol with two stones already on the board, handicapping the AI. “Even with a two-stone advantage, I feel like I will lose the first game to HanDol,” Lee said. “These days, I don't follow Go news. I wanted to play comfortably against HanDol as I have already retired, though I will do my best.”

Artikel ini pertama kali tayang di Motherboard