Shulou(Shulou.com)06/02 Report--

Https://www.toutiao.com/a6688958820816781836/

Since scientists proposed "AI Winter" in the face of the dilemma of AI research in the 1980s, the discussion about the "artificial intelligence" bubble has never stopped. Not long ago, an article called "AIWinter Is Well On Its Way" published by Filip Piekniewski, an American computer vision expert, caused extreme panic among entrepreneurs in the entire AI industry, including face recognition, intelligent robots, intelligent speech recognition and so on.

01

2017 is known as the first year of industrial application of artificial intelligence. KPMG's quarterly analysis of venture capital trends released in October 2018 shows that the technology industry has always dominated the global venture capital trend from 2013 to the third quarter of 2018 (figure 1). And Asian power has risen rapidly in recent years, on a par with American investors (figure 2).

Figure 1 Global Venture Capital Trends report for the third quarter of 2018-Investment Module data (data from KPMG)

Figure 2 Global Venture Capital Trends report for the third quarter of 2018-Investment Regional data (data from KPMG)

"Asia will continue to see a lot of venture capital activity in the coming quarters," said Egidio Zarrella, head of customer consulting and innovation at KPMG in China. "of all the investment areas, artificial intelligence is the most prominent in the eyes of venture investors." However, since 2013, China has gradually increased its investment in AI, the policy dividend has led the influx of capital, and China's investment in artificial intelligence has shown a rocket growth trend (figure 3).

Figure 3 growth trend of investment between China and the United States in 2007-2017

02

In recent years, computer vision and image technology, namely face recognition, has been the highest frequency of investment in the field of AI and the most promising direction for investors (figure 4). As the underlying technology of artificial intelligence, this technology has been studied for a long time, and the technology is more mature. Facial scanning payment, facial scanning unlocking, facial scanning ride, facial scanning authentication and other services are widely used, especially in finance, justice, public security, government, factory, education, medical and other fields, in which the field of security has become the main battlefield of face recognition technology.

Fig. 4 Frequency of input in AI field

More than 20 entrepreneurs showed off their artificial intelligence products at the China International Public Safety Expo in 2014. to this day, face recognition technology is still one of the highlights of the fair. The Chinese face recognition Market Research and Investment Prospect Forecast report from 2017 to 2022 (figure 5) points out that the direct market scale of China's face recognition industry exceeded 2.1 billion yuan last year, more than 2.7 billion yuan in 2018, and is expected to reach more than 5.3 billion yuan in 2021. Domestic face recognition applications are about to usher in a "blowout period".

Figure 5 Prediction of the size of the domestic face recognition market from 2016 to 2021 (data from China Industry Information Network)

03

The development trend of artificial intelligence proves the natural law of prosperity and decline. For example, the vertical industry of face recognition alone has gone from "blooming everywhere" to "bankruptcy" in just a few years. According to the death list of face recognition startups (figure 6), 55 smart companies created between 2015 and 2018 have declared bankruptcy, of which one in five entrepreneurs is a smart home. In the hotel industry, only two start-ups that develop "smart check-in" are still in business so far. Pierrot, author of the Centennial History of Silicon Valley, said: "due to the influx of capital into artificial intelligence in the short term, but companies, including Google, have not made a profit, there is a certain bubble in artificial intelligence." The emergence of the industry bubble is like a bolt from the blue, as if yesterday, market experts and researchers were still talking about the development of face recognition. Today, the end of entrepreneurs is coming.

Figure 6 death list of face recognition startups

This "death list" is a wake-up call for all face recognition entrepreneurs, revealing the huge gap between the future and reality, as well as reminding such entrepreneurs that it will be difficult. There are several reasons why AI's winter is approaching:

First: "pseudo-concept" is widely publicized in the market. Kaifu Lee once said on Weibo: "recently I met an underwear maker who also said he was an artificial intelligence enterprise. This is a very abnormal phenomenon." Blind investors and entrepreneurs eager for quick success and quick profit are over-touted in order to maximize their profits, and no matter what product is involved, they will put on a "face recognition" hat to create a gimmick as long as it involves a little bit of relevant algorithms. this impetuous phenomenon eventually led to investment flocking in and disappointed. For example, the face recognition early education machine that appeared on the market some time ago advertises that as long as the child is close to the machine, the machine will take the initiative to greet the child and guide the child to learn. Later, the researchers found that the learning machine had the same response to being close to people or things, and the child lost interest in the machine after three days.

Second, face recognition technology is not result-oriented. As far as the technology itself is concerned, "face recognition" is a technology, not a specific product or service, the focus is to be able to meet the rigid needs of individuals and solve the profit model of enterprises. Some entrepreneurs only focus on technological innovation and breakthroughs without systematically and comprehensively examining the market demand, resulting in a serious disconnection between technology and the market, which eventually leads to the collapse of the capital chain, such as the robot restaurant that appeared a few years ago. now most of them have withdrawn from the market.

Third: although there have been decades of international research on face recognition, and fruitful research results have been achieved, talents are still the biggest obstacle to the development of face recognition technology in China. Domestic face recognition technology started relatively late, the training mechanism is not perfect, coupled with the entry of capital in recent years, talents begin to be out of touch with capital. In response to this situation, entrepreneurs hire people at all costs. As a result, the company invests a lot of money in hiring, but most of the talents can not support the sustainable development of the industry.

04

AI's industry bubble is very much like the Internet bubble 18 years ago, but behind this industry reshuffle, the outcome is positive. First of all, this is a process of survival of the fittest and natural selection in the market. those start-ups with weak technology, "two feet in the air" and low matching with the market will be gradually eliminated in this baptism, leaving behind entrepreneurs who are really based on market research and development. Secondly, investors will return to rationality from the initial frenzy for face recognition technology, and have more consideration for the companies they invest in, prompting startups to constantly reflect on their own defects; moreover, talents will be precipitated in this industry reshuffle, more talents will be attracted to the research of artificial intelligence, the training mechanism will be constantly improved, the past sky-high salaries will not exist, and the situation of talent supply falling short of demand will be greatly alleviated. Finally, this industry bubble will give users a wide range of exposure to face recognition technology, and users' habits will gradually develop.

In the process of market elimination, some face recognition companies have been fully optimized, giving birth to a number of leading entrepreneurs with steady development. In the construction of the smart city, the FaceID online authentication system of Skyview Technology can complete dynamic face recognition, and its intelligent application has thousands of core customers, including Alibaba, Ant Financial Services Group, Foxconn, Lenovo, CITIC Bank and many other industry-level heads. Visual Weiye's face recognition bayonet system provides monitoring services for the entry and exit of people in all kinds of key places; in the financial field, the smart finance of Shangtang Technology has been applied in financial APP such as China Merchants Bank and Xianqianbao. Cloud's "facial scan withdrawal" and other services from science and technology have been adopted by more than 400 banks across the country. In the field of security, Haikangwei's face monitoring equipment has been used in important places such as the Bird's Nest, the Youth Olympic Village and the World Expo. Yitu's intelligent security platform provides technical support for China Customs, China Border Inspection, Huawei and so on. In the construction of smart hotel, Ali first put forward the concept of "unmanned hotel" in the future. in terms of intelligent check-in, Ruiwo Technology's "intelligent front desk" equipment successfully achieved "ten-second check-in and three-second check-out" and was put into operation in more than 100 hotels across the country.

Generally speaking, where there is a tuyere, there is a bubble, but the AI bubble is not appalling. On the contrary, it is like an industry test for artificial intelligence entrepreneurs. Entrepreneurs and investors must understand that smart applications must be market demand-oriented, and the fact is that by doing research, entrepreneurs who are still down-to-earth after the bursting of this bubble will have a much smoother future. We have to believe that Chinese entrepreneurs will keep rational thinking and catch up and seize a place in the AI field.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.