AlphaGo wins final game after neck-and-neck race

By Park Si-soo, Lee Han-soo

Google-developed AI program AlphaGo defeated top-ranked go player Lee Se-dol of Korea in their fifth and final game in Seoul, Tuesday, which ran neck-and-neck for five hours.

Lee lost after 280 moves; but this also indicated that Lee had seen through play patterns of AlphaGo to defend relatively well compared to the previous games.

The much-hyped tournament ended with 4-1 victory for AlphaGo, whose DeepMind developers will receive $1 million in prize money that they are donating to charities.

Lee had the upper hand during the beginning, leading Google DeepMind CEO and co-founder Demis Hassabis to post on his twitter feed: “AlphaGo made a bad mistake early in the game but now it is trying hard to claw it back … nail-biting.”

However, this did not last long as Lee allowed AlphaGo to catch up in the mid-game phase, when he played too passively.

Hassabis, a main developer of AlphaGo, hailed the victory.

“I’m kind of speechless,” he said during a post-match media briefing. “This was the most mind blowing game so far.” He said there were some mistakes made by AlphaGo in the early part of the game but in the end the program prevailed.

Lee looked disappointed.

“I’m disappointed that the challenge is over,” Lee said. “I couldn’t finish with a win. Even though I had the advantage in the beginning, I ended up losing. That means I still have lots of weaknesses.”