Google computer wins last video game against S. Korean Go master

datcouch April 14, 2016 20 likes

A Google-developed computer program had the last word Tuesday in its machine vs human challenge with South Korean Go grandmaster Lee Se-Dol, winning the last game for a sweeping 4-1 series success.

The win was vindication for AlphaGo's creators, DeepMind, who had promoted the program as a new form of expert system (AI) efficient in "intuitive" thought and with varied real-world applications.
"I'm type of without words, that was the most mind-blowing online game experience so far," stated DeepMind CEO Demis Hassabis.


"Early on ... it appeared that AlphaGo made quite a big error, but in the end it had the ability to get back into the match for an exceptionally close, extreme finish," Hassabis stated.
"We're simply type of stunned really," he included.


A noticeably upset Lee was left at the table with his head in his hands, after resigning at the end of a close-fought, five-hour fight with the AlphaGo program.
After Lee had actually handled to pull one back for mankind in game 4, AlphaGo was back on its finest harmful type, utilizing its deep neural networks to finally outsmart the 33-year-old in a stressful end online game.


"I began the match believing that I had a benefit, however that I was still beat appeared my shortcomings," Lee told press reporters afterwards.


The South Korean, who has 18 global titles under his belt and is extensively considered among the best Go gamers of the contemporary age, said he felt he had under-performed versus a formidable, however beatable opponent.


"I believe (AlphaGo) is still at a level that can be challenged by human beings and in that sense, like I said earlier, I feel a bit disappointed," Lee said.
"It is different, there's no doubt. Firstly, it’s not human. It required time for me to get utilized to its playing style," Lee stated.


"It's not shaken up psychologically and stays focused right until completion," he added.
Described as the "match of the century" by local media, the series was carefully enjoyed by 10s of millions of fans of the ancient parlor game mainly in East Asia in addition to AI scientists.
The "machine vs. human" component indicated the online games likewise made headlines around the globe, with AlphaGo's winning performance hailed as a watershed for the future of AI.
The most well-known AI triumph to this day came in 1997, when the IBM-developed supercomputer Deep Blue beat the then-world class chess champ Garry Kasparov.


AI holy Grail


Go, played for centuries mainly in Korea, Japan and China, had long remained the Holy Grail for AI designers due to its complexity and near-infinite number of possible configurations.
AlphaGo uses 2 sets of neural networks that allow it to crunch information in a more human-like fashion discarding millions of potential moves that human gamers would naturally understand were pointless.
It also uses algorithms that allow it to learn and improve from match play experience.
DeepMind states the innovation has far-ranging, real-world applications from making mobile phones smarter to health care, although Hassabis said they had no clear idea what the next step may be.
"We've truly just been focusing on the match until this moment," he told the post-match interview.
"We think there are many people things we can improve. Perhaps in the next few months we'll have the ability to reveal more concrete strategies," he added.


Before the final video game, South Korea's Go Association said it was awarding AlphaGo the greatest Go grandmaster rank of "ninth dan", booked for those whose ability at the ancient parlor game borders on "divinity.".

 

 

 

.