For AI, 2019 is a hard battle.
- Categories:Industry News
- Time of issue:2019-05-13 17:27
For AI, 2019 is a hard battle.
During the New Year's Day, I chatted with friends in academia. It detailed the development characteristics of the domestic academic wave since 2000: influenced by the LDA topic model in 2003, the laboratory has built a Graphical model to add hidden variables in the article. Hadoop 2004 Appearance, data heat has become the focus of various issues in various fields. Some people can publish new hot articles by rewriting the traditional algorithms in the MapReduce framework. After Spark comes out in 2010, Hadoop will be driven away, and then deep learning will appear. , and the heat of the topic model article has been reduced... from crazy to swearing, it is only a few years.
According to a report by the Dutch Elsevier Publishing Group, in the past 20 years, Chinese researchers have published about 134,000 research papers on artificial intelligence, and only 106,000 papers have been published in the United States. All people know that the current level of practical use, AI still has a way to go, but in theory, it is clear that China has taken an absolute leading position.
Behind the crazy academic world, there is no shortage of crazy capital markets. According to Zhang Xueli, deputy director of the China Information and Communication Institute of the Ministry of Industry and Information Technology, as of September 2018, there were 5,159 artificial intelligence enterprises in the world, with 1,122 (excluding Hong Kong, Macao and Taiwan) ranking second in China and 445 in Beijing. The city with the largest number of artificial intelligence companies in the world. In the first half of 2018, the global financing scale of artificial intelligence reached 43.5 billion US dollars, and China's scale reached 31.7 billion US dollars, accounting for more than three-quarters of the world. Another beautiful victory: China has only 3/4 of the world's financing by one-fifth of the world's enterprises, and investors are full of expectations for the Chinese AI field.
However, according to the report of the artificial intelligence report of the media, in 2017, China's AI startup company has accumulated more than 50 billion yuan of financing, but the top 100 companies in the commercial sector have accumulated revenues of less than 10 billion yuan. In the entire industry chain, more than 90% of AI companies are still in the loss stage, and most of them have annual operating income of less than 200 million.
In response to this phenomenon, Alibaba's former CEO Wei Zhe said: At present, the artificial intelligence bubble is huge, the media touts, the market is overheated. Many companies in the market claim to be "artificial intelligence" companies, but 90% of artificial intelligence companies are "pseudo-artificial intelligence."
The inability to commercialize technology is the key to artificial intelligence companies being labeled as “pseudo-artificial intelligence”. Part of the reason for the author's observation in the security market is that the company over-promotes the laboratory data, which makes the user fall in the actual experience. The other part is because the company blindly pursues imitation and lacks real innovation ability, including face recognition technology. Most notably, when product technology tends to be homogeneous, price wars will break out at any time.
Although the security market is large enough and scattered enough, fragmented demand can allow companies to access the market, but this may be different from the original intention of some CV companies. It is undeniable that artificial intelligence and deep learning have good development prospects in the fields of speech recognition, image recognition, and driverless driving. Together with the guidance of national policies, they will continue to prosper in the industry in the future. Popularization is less than the conventional reasoning of 5-10 years in the academic world. I believe that in the next ten years, it will be the battle for artificial intelligence. However, investors' expectations for the future and the enthusiasm for artificial intelligence have begun to return to the cooling-off period, which will not be as good as the future for the future development of CV companies.
Previous: Soft and flexible robots learn how to manipulate objects to achieve "everything can be used"