No one expected that artificial intelligence would be the first to refresh human three views in 2017.

AlphaGo, which became famous in March last year, turned into “Master” at the end of the year and played 60 rounds on the Internet, picking out almost all the top players and talented new players in the go field of China, Japan and South Korea.

Then, in the early morning of January 4 local time, DemisHassabis, founder and CEO of Google DeepMind, tweeted that “Master is AlphaGo”, officially announcing the creation of an upgraded version of AlphaGo.

The plot resembles Arnold Schwarzenegger’s “I will be back!”

Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include editing robots, writing robots and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.

AlphaGo has thrown down the gauntlet to a number of masters. “DeepMind, the team behind Master, said it wanted to play the last 10 games against the world’s once-dominant go players,” an official from the Korean Go Institute told CNBC. In this case, nie Weiping and Chang Hao, the ‘masters of chess,’ fought against Master.”

DeepMind also wants South Korea’s Cho Hoon-hyun (9th Dan) and Lee Chang-ho (9th Dan) to compete. However, Cho hoon-hyun, who has become a lawmaker of the ruling Saenuri Party, did not accept the invitation because of turmoil within the party.

The reason why Lee Chang-ho, another south Korean go champion, refused is more straightforward. He said he did not like playing weiqi on the Internet and refused to play it.

“If it’s just about winning or losing, after all, the AI doesn’t get tired or make any mistakes; This absolute advantage over humans, and the ability to train continuously according to certain rules, creates favorable conditions for ai to win, and it is not a surprise.” “Said South Korean female go player Yoon Young-min.

The speed of development is astounding

In March last year, AlphaGo shocked the world by beating Lee Se-dol, then a Professional Go player in South Korea. At that time, Yoon was convinced that Lee would win, saying, “If Lee loses, all of us who are baduk coaches will lose our jobs.”

Yoon Young-min runs a go training institute in Seoul, South Korea. Nearly a year later, China Business News visited Yin Yingmin, who was receiving parents who had come to take their children to learn Go.

“As a matter of fact, in the past year of the century-level man-machine war, I did not lose my job. On the contrary, many people paid attention to Go, and more parents began to pay attention to the promotion of go to children’s intellectual development. This may also be the benefits of man-machine war to go.” Yoon young-min said.

An official of the Korean Go Association also confirmed that the number of people asking to watch the game and applying to participate in the division of go increased significantly after the man-machine war last year.

For Master’s frequent success, Yin Yingmin seems to open up a lot. “Seeing so many top players from China, Japan and South Korea losing to AlphaGO, watching these idols fail one by one, I was still surprised and frustrated.” She said.

As for Master’s style and level of play, Yoon admitted that it was “much better than AlphaGO in previous man-machine battles, with a steady style and few mistakes”, adding that “The speed of development of AlphaGO’s ARTIFICIAL intelligence is really amazing”.

Mr. Hassabis said his team has been working to improve AlphaGo’s performance, testing unofficial go games with the new version several times in the past few days, thanks to accounts called Magister and Master, which run on servers called Tygem and FoxGo, respectively. Thanks also to everyone who watched. They were very satisfied with the result of the game. “More importantly, the new AlphaGo provides valuable lessons for the go community during the game.”

Hassabis singled out Gu Li, who failed at the chess game. “Gu Li posted on his blog that together, humans and AI will soon discover the secrets of deep learning,” he said.

“We look forward to the official test and will hold the full tournament again this year to explore the creativity of AlphaGo.” Hassabis said.

Can you still play go like this?

On December 29, 2016, Master appeared at eGO City for the first time and quickly won ten consecutive games against top professional players in China that night. At noon on The 30th, it won the first Korean player park Jung-hwan ninth Dan and the suspected First Chinese player Ke Jie.

“I’ve never seen such a move before. Can you play go like this?” ke said. “Humans have evolved over thousands of years of practice, and computers are telling us that humans have got it all wrong. I don t think anyone has even touched the truth of go.

On The night of January 4, in advance of regular matches against Shenzhen 9 Dan, Zhou Ruiyang 9 Dan and Gu Li, AlphaGo’s winning streak extended to 60 games, in accordance with advance plans, the unofficial match testing of the go program came to a temporary end.

At present, network fast chess is one of the main ways for professional players to practice. Because the thinking time for each step is very short, compared with the formal match with bonus, the opponents in fast chess are more prone to make mistakes. Therefore, the actual level of the upgraded AlphaGo still needs to be further tested by the formal match with rules.

Master shows a level of speed that human players can’t match, but given enough time to think, would the results be different?

The onslaught has begun

It is worth mentioning that on December 13 last year, a number of Google executives made a surprise visit to China, where they communicated with Nie Weiping, Ke Jie and others in the China Chess Academy and reached a cooperation agreement. “2017 is going to be a year of excitement for AlphaGo and the go community,” Hassabis said later at an award ceremony for Japan’s Journal Club. What people don’t realize is that the onslaught of ARTIFICIAL intelligence on this thousand-year-old art has already begun before the New Year arrives.

Zhu Zhi, founder and CEO of Sesame Technology, told China Business News: “This match showed AlphaGo’s better stability and confidence compared to the last three matches with Lee Sedol. In the field of Go, AI proved it was better, but that does not mean machines are intelligent. In fact, machines are better at dealing with these complex rational problems, so in many specific domains, the intelligence that machine learning brings will gradually change the traditional model.”

Whether the Master version has been upgraded or improved compared to AlphaGO, and what exactly has been optimized to make AlphaGO so powerful, the specific improvement plan is still not disclosed. Dai Wenyuan, founder and CEO of the fourth Paradigm, an artificial intelligence company, speculated that “after more than half a year of training, the amount of data should be increased a lot. Last time, we trained 8 million games of chess, and this time, it is estimated to be several hundred million. I think I have improved my training methods for my weakness in the fourth game (against Lee Se-dol).

Zhihu also sparked discussion on whether the Master should upgrade. A user identified as “founder of the Future Knowledge Graph” said that the reason for the improvement was that “the Master adjusts the algorithm and parameters based on AlphaGo, not how many games it has played. After all, the machine doesn’t really ‘learn’. It’s just reusing human-set algorithms to provide data for optimization parameters.”

However, Mr Dai said that the improved AlphaGo “may have some random or non-random disturbances in the game, that is, it chooses not the optimal position to play, tries new possibilities, and does not play against itself very clumsessly”.

Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.

KevinCurran, a senior member of the institute of electrical and electronics engineers (IEEE) and a professor of computer science at the university of Ulster in the UK, told China business news: “AlphaGo’s success surprised even its creators. We all knew AlphaGo was very advanced, but we didn’t expect it to be this good. Thus, the algorithm, which combines enhanced learning with tree search, effectively increases the ‘win rate’ that AlphaGo has over the best human go players. At the same time we should not forget that Google DeepMind team research the purpose of artificial intelligence technology, not only to develop a set of chess game tools, but in order to better in the future the application of artificial intelligence technology to the medical treatment, robot, driverless cars, agricultural, financial and other industry for the service of mankind. That’s what’s so amazing about technology today.”

The poet T.S. Eliot once said, “Last year’s words belong to last year’s words, and next year’s words await another voice.” This applies equally well to the rapid change in artificial intelligence.