China is strong in data and applications, and computing power is catching up

Source：Updated：2017-09-04 09:02:11

The United States is often considered the leader in artificial intelligence. But now China, the world's second-largest economy, has become the main contender. 
 
In its latest report on artificial intelligence, Goldman sachs says China has the resources and determination to build a smart economy based on artificial intelligence (AI) and deep learning in the coming years. In particular, China's Internet enterprises, led by BAT, are becoming an important driver of China's artificial intelligence development. 
 
In the development of artificial intelligence is the most important the four elements of "talent, data and computing power and infrastructure", China has now have the first three factors, in terms of computing power, although there is still dependent on foreign suppliers, but the domestic semiconductor industry is exciting progress. 
 
When it comes to the difference of artificial intelligence, a professor at the university of Pennsylvania, machine vision domain experts Shi Jianbo for the first financial journalist, said: "the application of artificial intelligence is based on open source platform, almost no gap between China and the United States." 
 
Push AI from the top down 
 
In November, Goldman sachs issued a 100-page report on artificial intelligence, saying that "the us is still the dominant force and China is growing at a rapid rate". 
 
After half a year, Goldman sachs report again hair, and with the rise of artificial intelligence of the People's Republic of China, this paper analyzes the present situation, the development of China's artificial intelligence and the future China will see more of the support of national and local level artificial intelligence development policy. 
 
Goldman sachs says that while the United States is widely regarded as the most advanced country in artificial intelligence, other countries are catching up. Big companies such as China's baidu, alibaba and tencent are examples. "China not only has resources but also has launched a top-down, ambitious plan to create a smart economy for artificial intelligence and machine learning in the coming years." 
 
The value of artificial intelligence is mainly reflected in four key areas: talent, data, infrastructure and computing power. China already has the first three factors, and while it still relies on foreign suppliers, China's domestic semiconductor industry is making exciting progress. 
 
In fact, artificial intelligence is ubiquitous, from the ability to predict the weather's smartphone app to the game of chess that supports complex algorithms that can beat human players. Big companies such as Google and Microsoft have invested heavily in artificial intelligence, hoping to expand the use of artificial intelligence. Machines are "fed" to large amounts of data and are "taught" a lot of specific tasks, enabling companies to create software that will learn and become smarter. 
 
In China, in July, the state council released a new generation of ai development plan (hereinafter called the plan) and set the development goal of China's artificial intelligence by 2030. By 2030, China's artificial intelligence will drive more than 1 trillion yuan of industry. 
 
The state council encourages the creation of open source computing platforms and trains more artificial intelligence professionals and researchers. The plan also said the government would invest in high-quality artificial intelligence projects and encourage private capital to invest. 
 
To this, a professor at the university of Pennsylvania, machine vision domain experts Shi Jianbo for the first financial journalist, said: "the application of artificial intelligence is based on open source platform, almost no gap between China and the United States, such as machine vision. In other ways, China is even expected to overtake the us, such as face-recognition technology based on human marker data. But the gap between China and the United States is larger in terms of the development of the underlying algorithms to systematized development, because it requires a lot of empirical work. 
 
Talent and infrastructure pave the way 
 
Shortage of talent reserve is a long-term problem in artificial intelligence development. In the process of training ai professionals, there is a lot of work to be done. 
 
In the us, tech giants are opening r&d LABS around the world to recruit talent. MIT, for example, began developing research projects and laboratories in the field of computer vision and robotics in 1963. 
 
MIT chancellor Eric in sen (Eric Grimson) to introduce the first financial reporter, a number of experts and professors involved in the MIT artificial intelligence programs, and set up the digital science laboratory and the media lab, etc., in the laboratory, under the joint innovation of mechanical arm sensors, unmanned technology to mature and gradually put into practice, the accumulation of decades of research process, driven by the wave of artificial intelligence, finally by value and commercial enterprises. 
 
Chinese companies have also begun to learn about the practices of American companies, and all three of BAT's giants have opened research and development centers in silicon valley. Earlier this year, baidu hired lu qi from Microsoft, and tencent also dug up yu dong from Microsoft, joining to strengthen the research and development of artificial intelligence of the two companies. 
 
When it comes to artificial intelligence infrastructure, most companies involved in ai r&d use open source platforms to attract talent and resources, Goldman said. Chinese companies follow the same rules. For example, baidu has developed its own artificial intelligence open source system, PaddlePaddle, which is a distributed, in-depth learning architecture. Baidu has also released another open source platform for autonomous driving, the Apollo program. 
 
Equally important is the algorithm and performance of ai, which relies on chip processors and computing power. Goldman sachs reported that while China's reliance on foreign chip suppliers was "very high", the progress made by local Chinese companies in the chip sector was "encouraging". 
 
Goldman sachs predicts that, over time, China will gradually reduce its dependence on foreign chip processors. For example, the kylin 970 processor, which is carried by China's latest Mate 10 series, is embedded with the Cambrian chip from the Chinese academy of sciences. 
 
Computation of Chinese academy of sciences in a public statement to the first financial journalist, said: "1 a kirin 970 chip integrates the Cambrian processor as the core of artificial intelligence processing units (NPU, namely network processor), phones local, real-time and efficient intelligent processing." It also argues that this is a big step forward in the production of smart processors for the Cambrian. 
 
But America still has a firm grip on its core technology. The trump administration will decide in the next two weeks whether to approve a deal to have a chinese-backed chip field. Last November Canyon Bridge, a firm, announced a $1.3 billion acquisition of U.S. chipmaker Lattice. But the deal was opposed by the federal government commission. The company is a chipmaker in the field of FPGA, the main competitor to syringes and Intel's Altera. 
 
BAT benefits most 
 
Goldman notes that China's large number of Internet users give Internet companies a huge advantage in producing data. And China's Internet giants tend to have a more comprehensive ecosystem that permeates every aspect of people's lives, and that will produce explosive data. 
 
According to the report, China now produces 13% of the global digital information content, and by 2020, this proportion will grow to 20% to 25%, equivalent to produce 9 to 10 Zettabyte (ZB, 1 ZB = 1 trillion GB), at that time, China will become the world's largest economy. 
 
Goldman sachs predicts that the initial beneficiaries of artificial intelligence will be BAT. That's because the three Chinese Internet giants have vast and unique databases that give them a unique advantage in artificial intelligence. 
 
Other businesses that benefit include e-commerce and Shared travel. Meituan, for example, reviews these companies that offer Internet services. Goldman sachs estimates that they can use big data analytics to generate the most efficient distribution routes in less than 100 milliseconds. Didi chuxing is using technology such as deep learning, human-computer interaction, machine vision and intelligent driving. Goldman sachs estimates that didi chuxing handled more than 4500TB of data, receiving more than 20 billion routes and more than 20 million daily orders. 
 
The smart voice industry is also a focus of the Goldman sachs report. According to the report, hkust is focused on speech recognition, which has the largest market in China's intelligent voice industry. 
 
Report also noted that some emerging startups, such as voice interaction and natural semantic recognition go ask and focus on machine vision and depth of Thomson learning science and technology, and unmanned aircraft maker xinjiang and the kind of person robot manufacturers optimal choice. 
 
Currently, voice and visual recognition technologies account for 60 percent and 12.5 percent of China's artificial intelligence market, according to iresearch, cited by Goldman sachs in an artificial intelligence report. In China, all companies involved and artificial intelligence, 71% focused on the development and application, the rest of the company focus on algorithms, among them, 55% of the company's research of computer vision, 13% of the study of natural language processing, 9% on machine learning. 
 
China's artificial intelligence market will grow from 1.2 billion yuan in 2015 to 9.1 billion yuan in 2020, according to iresearch.