Google's AlphaGo tasted defeat for the first time at the hands of human Lee Se-dol on Sunday when the Korean grandmaster defeated the AI program in a match which lasted almost 5 hours. This was the fourth match of the five-match series being held in Seoul. Earlier the Google program had created a history by defeating the 18-times international champion 3-0 and winning the series.
This is the first human defeat ever for the intuitive Go playing program as it had defeated the European champion Fan Hui 5-0 the last year. The fifth and the final game is to be played on Tuesday. Go is an ancient Chinese game played with black and white counters on a 19 x 19 board. The players place the counters on intersections to cover the most territory. The game involves a lot of calculation and logical intuition.
The AI program uses two set of deep neural networks to predict many possible moves. Many experts deemed its strategy unorthodox for a Go player but eventually it came through. But on Sunday, the program which seemed invincible finally made a mistake. Initially thought that the mistake was made around move 87, it was later revealed that the computer realized the error was made around move 79.
Demis Hassabis, the lead of the Deepmind team who has developed AlphaGo congratulated Se-dol. He said, "While the computers have mastered all the other games, this Chinese game is a tough challenge. And this is why we are here". Earlier when AlphaGo defeated the grandmaster 3-0 he said, "We are a bit of stunned and speechless. But we don't see this as a loss for the humanity, we will be able to solve a lot of problems with the approach AlphaGo is using".
See-dol said on the victory, "I am very happy. I wouldn't trade it for the anything in the world". Before the series he had predicted that he would win easily but that turned out to be wrong. The champion said, "AlphaGo seems to be weaker playing with the black counters. This might be a bug or a genuine weakness, I can't tell".
You can see all the previous matches on Deepmind's YouTube channel.
![]()
Like Us On Facebook |
![]()
Follow Us On Twitter |
![]()
Contact HuffPost India
This is the first human defeat ever for the intuitive Go playing program as it had defeated the European champion Fan Hui 5-0 the last year. The fifth and the final game is to be played on Tuesday. Go is an ancient Chinese game played with black and white counters on a 19 x 19 board. The players place the counters on intersections to cover the most territory. The game involves a lot of calculation and logical intuition.
The AI program uses two set of deep neural networks to predict many possible moves. Many experts deemed its strategy unorthodox for a Go player but eventually it came through. But on Sunday, the program which seemed invincible finally made a mistake. Initially thought that the mistake was made around move 87, it was later revealed that the computer realized the error was made around move 79.
Lee Sedol wins game 4!!! Congratulations! He was too good for us today and pressured #AlphaGo into a mistake that it couldn’t recover from
— Demis Hassabis (@demishassabis) March 13, 2016
Demis Hassabis, the lead of the Deepmind team who has developed AlphaGo congratulated Se-dol. He said, "While the computers have mastered all the other games, this Chinese game is a tough challenge. And this is why we are here". Earlier when AlphaGo defeated the grandmaster 3-0 he said, "We are a bit of stunned and speechless. But we don't see this as a loss for the humanity, we will be able to solve a lot of problems with the approach AlphaGo is using".
See-dol said on the victory, "I am very happy. I wouldn't trade it for the anything in the world". Before the series he had predicted that he would win easily but that turned out to be wrong. The champion said, "AlphaGo seems to be weaker playing with the black counters. This might be a bug or a genuine weakness, I can't tell".
You can see all the previous matches on Deepmind's YouTube channel.
Like Us On Facebook | 
Follow Us On Twitter | 
Contact HuffPost India