When the machine wins, the machine wins. This is the inevitable result of technological development. But, “it’s a long way from ai threatening humanity.”

Helen

Artificial intelligence (AI) has successfully challenged the human brain again after AlphaGo’s Master beat the best go players with a record of 60 wins, 0 losses and 1 draw.

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.

In the first “man-machine war” of “The Brain” recently, Baidu artificial intelligence defeated Wang Feng in the task of “cross-age identification” with the score of 3∶2. In the field of “face recognition”, robot “small degree” record is outstanding.

In the previous, from Deep Blue to AlphaGo, the focus of the war includes memory ability, logical reasoning, computing ability; In addition to these, ai will also challenge the sensory talents of hearing and vision, which have evolved over millions of years.

Lin Yuanqing, director of Baidu’s Deep Learning Lab (IDL), told China Business News that due to the program’s time limit, the entire process of man-machine war has been cut from the TV program. The first part of the current broadcast of the man-machine battle (” Bee Girl “) is complete, with Xiao Du and Wang Feng tied at 2 to 1. In fact, the second part of the game compared three games, the first game is both wrong, the second game both right, the third game Wang Feng wrong, small degree of pair, the battle was 2∶1. The score of the first part was only one point, and the score of the second part was two points, so the whole match was decided in the last game. From the point of view of scientific experiment, a total of 5 games, among them “small degree” of the 4 bureau, Wang Feng right 2 bureau.

Wei Kunlin, a professor at the School of Psychology and Cognitive Sciences at Peking University, said artificial intelligence has reached a high level of accuracy in face recognition, and the next step should be to improve the efficiency and energy consumption. “If this era, artificial intelligence is coming so fast, what should we do to make better use of our abilities, to develop better, to survive better?”

Micro plastic surgery, twins are not afraid

Baidu artificial intelligence first “man-machine war” opponent, is “90 empress” player Wang Feng, his memory cognitive ability is top class. He has memorized 500 numbers in 5 minutes, 2660 numbers in 1 hour and 300 English numbers by listening. He has broken three world records, won the “World Memory Champion” of the World Mental Championship for many times, and broke the world record of fast memory poker.

This time, the competition is the cross-age version of face recognition. According to the rules of the competition, in the first round, Zhang Ziyi selected 2 difficult photos from the 20 childhood photos of the members of the Bee Girl team. The contestants matched the selected childhood photos with the adult girls present through the dynamic video performance of only three seconds on average. In the second round, they observed an audience over 30 years old and then picked her out of 30 group photos taken in primary school.

Professor Andrew Ng, baidu’s chief scientist, admitted that the tasks in the competition were very difficult for machines and relatively easy for humans. Machines have no intuition and no long evolutionary history and can only learn by analyzing data. “There’s not a lot of data to analyze to identify plastic surgery, heavy makeup, or photos that span more than a decade.”

The technical difficulty of these two rounds of competition lies in the fact that there are a large number of bee girls team members and everyone performs makeup on the field, which is not excluded from the interference of micro-plastic surgery, wearing contact lenses and other factors. In addition, the selection of childhood photos are in the range of 0 to 4 years old, and the present age span is larger. In the second round, the sample size is large, and 30 group photos need to find the corresponding person in 1000~2000 faces, and the age span is also covered in the 80s, 90s and other age groups. At the same time, there are real-time photo transmission, live camera to capture face image shaking, lighting interference and other factors, which will affect the recognition accuracy of artificial intelligence.

After a few minutes of breath-holding, the owner of the first childhood photo was easily found by both parties. However, when it came to identifying the owner of the second childhood photo, Baidu Brain came up with two different answers with a probability of one in 10,000. Ng explained that this was because the probability of the two answers was so close that it was highly likely that there were twins present!

Finally, the robot with probability to find the answer, this round, Wang Feng narrowly lost; Both sides were able to pinpoint the right person in more than 1,000 photos.

Interestingly, after the match, Lin yuanqing was asked by the media: “If AlphaGo, which has beaten almost all go masters, had a match with Baidu Brain, who would win?”

The answer given by Lin Yuanqing is that the direction chosen by Google and Baidu, especially the direction of “man-machine war”, is very different. “When AlphaGo came out last year, we were putting a lot of effort into driverless cars. Baidu didn’t have a go program.”

Artificial intelligence is not perfect

From AlphaGo, which beat almost all go masters, to Baidu’s brain, which beat the human brain in the first man-machine war, will artificial intelligence overturn the human world in the future?

In luo Zhenyu’s eyes, ARTIFICIAL intelligence is not only an extension of human beings, but also a replacement of human beings. Citing Yuval Harari’s new book, “A Brief History of the Future,” he said, “There may be a kind of useless person in the future. If ai were powerful enough to be exploited and enslaved without complaint, some humans might not even be worth being exploited. What is looming is the disappearance of a range of professions.”

In Wei kunlin’s view, if the machine wins, the machine wins, which is the inevitable result of technological development. Sooner or later that day will come, but sooner or later.

“Every time a new technology comes along, people panic: cars panic, trains panic, computers panic… This is the ultimate panic. The emergence of ARTIFICIAL intelligence could threaten many general intellectual activities (including many white-collar workers) and even the jobs of professionals (including doctors in some fields). But some jobs die and new jobs are created.” Wei Kunlin said.

Lin yuanqing, however, believes that artificial intelligence may replace some low-end jobs, such as assembly line robots in car factories. “Our view of AI as a threat to human beings is far away.”

At present, the industrial applications of artificial intelligence face recognition include but not limited to remote pre-credit and face recognition gates in the financial field.

In the face recognition arena, however, AI isn’t perfect right now.

Lin Yuanqing told reporters that the deep learning method used by Baidu Brain can learn commonalities from massive amounts of data, but if a person has a mole growing around his mouth, machine learning is difficult, because it is not common. For those more unique, personalized characteristics, ai has not made good use of.

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.

Another example is cat recognition. Even a three-year-old child can recognize cats with different poses and colors. For a machine to do accurate recognition, it needs a very large amount of data training, maybe thousands of data; The results of the training may not be correct, and the next time you encounter a cat with a different posture, you may be wrong again.

“Although our AI technology has made great progress in the past few years, recognition is still a very difficult thing to do.”