Artificial intelligence: opponents in high hands?

In 2017, artificial intelligence will continue to lead the trend of technology, artificial intelligence + will also be like 2016 VR+.

In March 2016, artificial intelligence AlphaGo once defeated the professional Korean Go player Li Shishi in South Korea, shocking the world. Just after the 2017 New Year, Master, the latest version of AlphaGo, has defeated Ke Jie, Chang Hao, Shi Yue and South Korean world champions Park Ting-soo, Jiang Dong-run and others in the game network and the wild fox network, winning the 60-win record. In the chess field, the Master defeated almost all the top players in the Chinese, Japanese and Korean Go game.

Some people say that Go is the last barrier of mankind. Successive failures begin to make people feel sad, panic and anxious. If artificial intelligence really surpasses humanity in wisdom, will humans become slaves to AI? Is the last line of defense for humans a Go? Is the ultimate enemy of mankind AI? For various conspiracy theories in the market, the author will share his views.

If artificial intelligence defeats the master of Go, is it also in other fields? In the geek departure column of CCTV 2, the security inspection robot with powerful artificial intelligence in the face recognition process, due to the early stage The collected expression images are limited, and the host has deceived the robot through disguise. Secondly, researchers at the National Institute of Informatics (NII) announced that they have abandoned the plan to get the "Torobo-kun" robot into the University of Tokyo because artificial intelligence has a foundation in face recognition and speech recognition, but on humans. The reading comprehension of the language seems to be underperforming. Torobo-kun has been a "re-reading student" for four years and still cannot enter the University of Tokyo.

Based on the above cases, the author believes that artificial intelligence is still at the stage of top students in some subdivision applications, and even far behind the school. If artificial intelligence is to cross the school, the following three conditions must be met.

1, huge big data, as many scientists. The super-cognitive learning of artificial intelligence comes from Google's powerful big data system. With the updating of knowledge digitization, more and more knowledge is digitized and digitized. If a person has 8 hours of reading time, then artificial intelligence is 24 hours of reading. Enrich your knowledge through continuous learning throughout the day,

2, a powerful system algorithm, often said that learning can not read the dead book, we must learn to give the opposite. In artificial intelligence, if artificial intelligence organizes the data on the Internet, it is only an advanced manpower, not intelligent. Alpha's ability to defeat Go Masters and Texas Hold'em professional players is a cutting-edge algorithm. Some experts pointed out that defeating Li Shishi does not mean conquering Go, but an algorithmic victory.

3, excellent hardware configuration, processor chip design and process breakthroughs drive the development of electronic information industry is an indisputable fact, but for the application of artificial intelligence, the improvement of the processor alone is not enough. Internet giants such as Google and Baidu and Intel and Nvidia are doing complementary technologies in the field of artificial intelligence. Like Google, it is more about cloud and data center processing. The chip for artificial intelligence is a large-scale industry of artificial intelligence. The key is also the bottleneck and core cause of the inability of large-scale industrialization of artificial intelligence today. Overall, the current development of artificial intelligence industry is at an early stage. At present, the level of chips and hardware of artificial intelligence is equivalent to the hardware level before the invention of personal computers 50 or 60 years ago. With the personal computer, the Internet and mobile Internet are available.

According to foreign media reports, Japanese Mitsubishi Research Institute scholars believe that Japan's active use of artificial intelligence technology will lead to a reduction of 2.4 million jobs in Japan after 13 years. Using robots and artificial intelligence technology will create 5 million jobs by 2030. At the same time, because of the replacement of labor by robots, the manpower demand in the traditional field will be reduced by 7.4 million. Therefore, 2.4 million Japanese will lose their jobs.

Robots will replace manual labor in several industries or assist humans in completing their tasks. Dr. Kai-Fu Lee predicts that about 90% of people working in translation, news reporting, assistants, security, sales, customer service, transactions, accounting, drivers, housekeeping, etc. will be replaced by artificial intelligence in whole or in part in the next decade. At the same time, some new job opportunities will emerge, and some of the traditional work will be transformed into a new type of work assisted by artificial intelligence.

In fact, in the history of mankind for thousands of years, it is difficult to eliminate human beings because of the upgrading of a certain technology, from the firewood in ancient times, to the industrial revolution in the industrial age, and the Internet revolution in the civilized era. Behind every technology, people are more brave enough to face nature and challenges.

The author will outline the following two aspects, artificial intelligence is an assistant, not an opponent.

1. Simplify the problem. For example, in the translation industry, Microsoft has released a universal translator that supports multi-person multi-language communication in the western United States - Microsoft Translator. With Microsoft Translator, everyone can use their own language on their own smart devices. Communicate with anyone, this is the world's first universal universal translator. But in the face of local accents and dialects, artificial intelligence can be helpless. It will only keep reminding you of "Mandarin". In such an environment, many times, human assistance is needed to complete the work. But in everyday life, artificial intelligence can be generalized and complicated. Quickly assist humans with complete system algorithms.

2, improve office efficiency, such as the accounting industry, whether it is accounting, financial, or intermediary practitioners, can not help but feel that there is a large amount of repetitive work in daily work. Invoices are repeatedly reviewed, various forms are repeatedly prepared, and various data are repeatedly checked. It’s really not erupting in repetition, just like everyone in the repetition. The emergence of artificial intelligence can allow everyone in the industry to jump out of the original mechanical, repetitive work, to do more valuable things, rather than eliminating accountants.

Although artificial intelligence will spawn several billion-dollar or even trillion-dollar industries, governments across the country are actively deploying artificial intelligence industries. For example, Chengdu Jingrong International Plaza and Chengdu Silicon Valley International Incubator are recruiting artificial intelligence entrepreneurs. However, for the current artificial intelligence industry, it is necessary to improve the following aspects.

1. The technical framework, platform, tools and services that can be reused and standardized are not yet mature.

Although deep learning frameworks such as TensorFlow and MXNet have been adopted by tens of thousands of R&D teams, the number of related open source projects is increasing rapidly, but a complete artificial intelligence ecosystem is necessary, from chips, buses, platforms, architectures to frameworks. , application models, assessment tools, visualization tools, modularization and standardization of cloud services, it will take three years or more to really mature.

Second, the talent gap is huge, and the talent structure is unbalanced.

According to LinkedIn, there are currently about 250,000 artificial intelligence professionals worldwide, of which about one-third are in the United States. This level of talent pool is far from meeting the macro demand for rapid and steady growth of artificial intelligence in the vertical and consumer markets in the next few years. The contradiction between talent supply and demand is significant, and the value of advanced algorithm engineers, researchers and scientists continues to rise. In terms of talent structure, the proportion of high-end talents, backbones and basic talents is far from optimal.

Through online recruitment search artificial intelligence, the main talent gap artificial intelligence engineer, skills need to master human-computer interaction, deep learning, NLP and other fields of cutting-edge technology, with big data mining and analysis capabilities. Secondly, artificial intelligence algorithm engineers need to use the latest deep learning and machine learning algorithms, image recognition, semantic understanding, speech recognition and other algorithms.

Third, the difficulty of starting a business is relatively high, and the early startup team needs more support.

Compared with the Internet era and the entrepreneurship of the mobile Internet era, the artificial intelligence startup team faces many new challenges. For example, relying on high-level talents, scientists and entrepreneurs have less business experience, high-quality big data is difficult to obtain, deep learning units and computing clusters are very expensive.

The entrepreneurial team must have the knowledge and ability of multiple disciplines, or face a variety of uneven development platforms in the market. Entrepreneurship in the field of artificial intelligence is undoubtedly explored in markets unknown to the world. For this kind of exploration, more is ignorance and unknown.

In 2017, artificial intelligence will continue to lead the trend of technology, artificial intelligence + will also be like 2016 VR+. A large number of entrepreneurs and capital influx, when the tide quits, only know who's body floats in the water.