Ke Jie's first defeat, the artificial intelligence represented by AlphaGo, has to rule the earth?

Ke Jie responded to the defeat in the first battle and lost AlphaGo. Ke Jie believes that the main reason is that he has not found its weakness. "It really is too good. Many places are worth learning. AlphaGo lets us understand that no chess can't be played. It Increasingly, we are changing our initial understanding of Go.” Ke Jie said that if AlphaGo is viewed as a person, it is exactly two people this year and last year, because artificial intelligence is evolving.

From 10:30 in the morning to 3 in the afternoon, Ke Jie battled Black and AlphaGo for more than four hours. He started with a small three-three, but AlphaGo was not affected, and continued to suppress Ke Jie with super computing power. Although Ke Jie did his best at the palace stage, he did not find a chance to reverse. After the battle was completed, the black board was 6 eyes. According to the Chinese rules, Ke Jie lost with 1/4.

It is undeniable that this game Ke Jie has played very well, but still can not stop the progress of AlphaGo. Although it is the smallest gap in the end, from the process point of view, Ke Jie has not had a good chance to win. What is rare is that Ke Jie put away his usual pride before the game. "I think that we humans have struggled hard if they have worked hard. Even if they lose, there is no regret." As a result, they lost the game with the least disadvantage in the Chinese Go rules. It has been successful for Ke Jie.

"Even if the Alpha dog defeated Li Shishi, it won't win me." Once, Ke Jie treated AlphaGo like this, but from the end of last year to the beginning of this year, the first person in the world of Go suffered a losing streak when facing the new version of AlphaGo Master. . In today's Wuzhen, in the world-renowned man-machine war 2.0 battle, Ke Jie is still defeated, but he is only a quarter of a small negative, still have a chance to win?

It should be pointed out that, unlike last year's man-machine war 1.0, the time spent on the competition was Li Shishi's time spent on AlphaGo for 2 hours per side, 5 minutes per minute, and this time it was 3 hours per minute, 1 minute. Five counts of readings, this change is naturally to provide greater benefits to humans, but this still can not help Ke Jie beat AlphaGo.

"It really is going too well. Many places are worth learning. AlphaGo lets us understand that there is no chess that can't be played. It is changing more and more of our initial understanding of Go." Ke Jie said if AlphaGo is a person, and it is exactly two people this year and last year.

On the day before the game, Ke Jie said on Weibo that this was his last three human-machine battles. For the next two games, he said: "AlphaGo's weaknesses have not been found yet, but I have always had confidence in myself. I think it is the most honored thing in my life to be able to confront AlphaGo, so I will go all out to the bottom. Two chess games, cherish this opportunity."

In the five-day Go Summit, in addition to the three battles between AlphaGo and Ke Jie, Coulee and Lian Xiao will be paired with AlphaGo, and the five top Chinese professional players will form a battle group. Collective intelligence against artificial intelligence. Ke Jie will debut again at 10:30 tomorrow morning against AlphaGo.

According to Hong Kong's "Economic Daily" reported on May 23, according to the data given by the AI ​​God operator of Tsinghua University, the two sides played chess to 106 hands, Ke Jie winning rate continued to increase to 45.91, behind only 7 orders. To 108 hands, Ke Jie wins 45.63.

Compared with Li Shishi, the Korean chess player who played AlphaGo last year, Li Shishi completed 186 hands in 4 and a half hours in the first game. His winning percentage has remained at 40% and once rose to around 47%, but then fell to only two after 116 hands. Into more.

According to preliminary estimates, Ke Jie's performance may be better than Li Shishi last year.

Of course, these reports are the sayings of Hong Kong media. In any case, we all believe that people will win the day, but in some computing and functional aspects, human beings are indeed better than machines, but machines are serving human beings, just like Chinese Go Jie and the Google artificial intelligence system AlphaGo's man-machine Go war, despite the defeat, but compared with the South Korean player Li Shishi who challenged AlphaGo last year, Ke Jie's performance is even worse.

Wang Weinan, chairman of the China Go Association, believes that it is still very interesting so far. Ke Jie is well prepared and the preparations are still very detailed.

When Korean chess player Jin Tingxian commented on Ke Jie’s performance today, Ke Jie and humans always looked calm. But today, the face is full of blood, it seems very nervous.

AlphaGo challenges the world's number one Go national team Ke Jie nine paragraphs. For this matter, the innovation workshop Li Kaifu commented that the victory rate of this matchup is almost zero.

Kai-Fu Lee said that AlphaGo and Li Shishi's man-machine war is the first time, and there may be suspense, but now the evolution of AlphaGo and Ke Jie's man-machine war, no longer have any other possibilities. Li Kaifu pointed out that the game between AlphaGo and Ke Jie is not meaningless, but has lost its focus on the level of scientific value. He appealed that since AlphaGo and the master of cold flutter have made the machine "unbeatable" in the game field no longer suspense, it is better to put more energy into the commercial application level of artificial intelligence.

Another point of view is this: If artificial intelligence is like a creature, why is this creature more persistent than humans playing chess?

It is difficult to tell who was the first to teach computers to play chess. Claude Shannon, the greatest computer science expert of the twentieth century, published a paper in 1949, "Programming to Realize Computer (International) Chess."

Another scholar, Alan Turing, who was also involved in the history of computer development, wrote the first computer chess program in 1952. At that time, the computing power of the computer could not support the operation of this program. Turing used pen and paper. Performing a simulation check against a human colleague, it takes half an hour to take each step and finally loses.

If AlphaGo beats Ke Jie, it means: from learning to play chess to winning the game in the field of board games, artificial intelligence has been waiting for 50 years. If it loses, it will continue to evolve. In fact, as early as the 1950s, 70s, 90s, and early 21st century, it was predicted that "artificial intelligence will soon rule humanity."

Be aware that before deep learning occurs, single-layer neural network algorithms can only make simple linear decisions. In order to solve the nonlinear judgment, the multi-layer neural network invented by the researchers requires almost impossible calculations in training. Until the above-mentioned parameters appear one layer after another, the "backpropagation" mechanism is automatically transmitted, which makes deep learning a practical technology.

Every "neuron" in a multi-layer neural network is as simple and elegant as the neurons in our brain, but at least to some extent simulates our brain's perception and judgment of the world. This allows artificial intelligence with deep learning to be described as a "child". The advancement of artificial intelligence will come from two aspects: on the one hand, the explosive algorithm improvement, on the other hand, the progressive power increase.

Artificial intelligence has been continuously improving, reviewing the progress of artificial intelligence, although time is not certain, it is not difficult to come to the conclusion that artificial intelligence will eventually defeat humanity.

Human research on algorithms is more like the discovery of the objective existence of the universe. Although it still needs to accumulate, try and the wisdom of scientists, its impact on artificial intelligence is abrupt. Just as we can divide physics into Newton's discovery of gravity and after discovering gravity.

Multi-layer neural network algorithms may not be the ultimate algorithm for true advanced artificial intelligence in the future. This means that even artificial intelligence in this era, represented by AlphaGo, may always be the current mental retardation in real life. But it does not mean that the development of artificial intelligence will be stagnant.

Unlike algorithms, the advancement of computing power is a test of engineering precision and industrialization level. Although Moore's Law has been ineffective, leading-edge computing devices such as quantum computers and bio-computers may continue to make progress in computing power.

Around AlphaGo, the technologies behind it include image processing, big data analysis, etc. These technologies are still in use in other fields at an early stage of exploration. They are only used in certain areas of AlphaGo research, but will certainly be in multiple fields in the future. Promote related technologies.

Demis Hassabis, the father of AlphaGo

Demis Hassabis, the father of AlphaGo, said that the current information overload and system redundancy are huge challenges for humans. I hope to use AI to find the meta solution. "Our goal is to achieve 'artificial intelligence scientists' or 'artificial intelligence-assisted science'."

“Artificial intelligence, like all powerful new technologies, benefits humanity in the ethics and responsibility constraints,” Demis Hassabis said. This means that artificial intelligence should be applied to science, pharmaceuticals and other fields, not to research and development of weapons and war; in addition, artificial intelligence can not be used by only a few companies, but shared by all human beings.

We did find that AlphaGo has shown creativity, and in a certain field it can even imitate human intuition. However, Xiao Bian believes that the probability of scenes dominated by robots in science fiction is very small. Since it is expected that this will happen, human beings will not use tools and restrictions while developing artificial intelligence. There must be a win-win strategy to evade.

43inch Horizontal Query Machine Integrated

10,4+64G,X86 i3 intelcore CPU;

Lcd Capacitive Advertising Display Screen,Business Guidance All-In-One Touch Screen,Lcd Touch Screen Query All-In-One,Query Machine Integrated Monitor

Jumei Video(Shenzhen)Co.,Ltd , https://www.jmsxdisplay.com