Ray Kurzweil predicted in his book "The Singularity is Near" that the "singularity" where computer intelligence will completely surpass humans will arrive in 2045. Looking back from this not-too-distant time point, the current intelligent application should be in the "prelude" stage of comprehensive advancement and multi-point explosion. This is also true from the facts. Intelligent transformation in different fields such as finance, medical care, transportation, and industrial manufacturing has been rapidly unfolding in a few years, and the underlying driving force of this process is the accumulation and development of big data.
Cultivate intelligent applications in the soil of big data
Artificial intelligence applications have four key elements: algorithms, computing power, data and application scenarios. For artificial intelligence, the development of big data technology is the basis for the exploration and progress of artificial intelligence technology. Since the 1990s, the development of Internet technology and high-speed computers has led to the explosive growth of information. As a result, innovative research on big data technology has achieved leapfrog development. According to IBM's summary, big data has four characteristics: large amount, high speed, diversity and low value density. The combination of large amount and low value density undoubtedly amplifies the difficulty of big data in the process of value mining.
On the other hand, in 2006, Jeffrey Hinton and others proposed the concept of deep learning, ushering in a new wave of development of human intelligence. In recent years, algorithms such as machine learning and deep learning have been widely used in the field of artificial intelligence, and have also promoted the development of computer vision, natural language processing and other fields.
According to Gong Ke, executive director of the China New Generation Artificial Intelligence Development Strategy Research Institute, the significance of data to artificial intelligence is: data collection is the basis of deep learning, and algorithm training and verification are inseparable. Collection of data. As a result, big data and artificial intelligence have established a close connection through the bridge of deep learning.
Taking daily online shopping scenarios as an example, a large amount of behavioral data will be generated every time a consumer clicks on a shopping website and even the time they stay on different web pages. For platforms, discovering consumers’ interests, hobbies and purchasing habits from this data is their most urgent need. If we only rely on the experience and judgment of staff to obtain the required information from these data, it will be difficult to guarantee accuracy and timeliness despite the heavy workload. At this time, it is necessary to take advantage of artificial intelligence, design corresponding machine learning algorithms, and conduct training on a large amount of online shopping behavior data. Subsequently, the algorithm will be continuously optimized based on the feedback received, and finally the product push plan and website design ideas that best fit the consumer psychology will be found to achieve the goal of improving the platform's transaction efficiency.
In the process of epidemic prevention and control, sensors such as cameras and infrared detectors are deployed in various scenes in daily life. With the orderly resumption of work, production and classes, people's travel activities have become more frequent. These sensors will generate massive amounts of image and text data every day. It is obviously "impossible to rely on human power to analyze these data to obtain epidemic-related information." mission". In recent years, computer vision technology has been fully trained in scenarios such as smart security and environmental monitoring. Therefore, in the face of the epidemic, algorithms can be quickly customized and combined with background databases to capture parameters such as body temperature, itinerary, and vaccination status in a timely and accurate manner. , to provide guarantee for epidemic prevention and control.
In addition, with the advancement of my country's information infrastructure construction and the leading development of 5G networks, big data will usher in continued explosive growth, bringing benefits to the development of artificial intelligence technology. According to the "Digital China Development Report (2020)" released by the Cyberspace Administration of China, my country has built the world's largest optical fiber network and 4G network; the speed and scale of 5G network construction ranks first in the world, and the number of 5G base stations built reaches 71.8 Ten thousand, the number of 5G terminal connections exceeds 200 million; mobile Internet user access traffic increased from 4.19 billion GB at the end of 2015 to 165.6 billion GB in 2020.
Combining these current situations, we might as well compare big data to a vast and thick land, but the fertile soil is also full of gravel and rocks, and artificial intelligence is like a farming tool. It not only improves the quality of the land, but also optimizes its own working methods and mechanisms, thereby cultivating intelligent applications with diverse functions.
From big data applications to intelligent applications
According to the principle of technological evolution, big data applications are equivalent to intelligent applications to some extent.
But in reality, the seemingly natural development process from big data applications to truly intelligent applications is actually the direction that one big data company after another has succeeded in, and has been repeatedly honed by practice.
2013 is called the first year of big data. This is the starting point of industrial development marked by the rapid increase in the number of big data companies and the rapid influx of capital. Prior to this, big data and artificial intelligence at the technical level had actually made some leaps and bounds. For example, the deep learning algorithm supported by artificial intelligence technology has been proposed as early as 2006.
After more than ten years of development, from the big data boom to the artificial intelligence boom, from the earliest big data companies to the artificial intelligence companies that are still alive today. What we see is no longer the rise of a new industry as previously expected, but big data companies and traditional industries exploring new industry solutions together.
It can be seen that many of the artificial intelligence companies currently developed are developed from big data companies. An interesting phenomenon is that more and more big data companies in the past tend to introduce themselves as an intelligent company, and some companies have even changed their names and added intelligence to the company name to highlight "intelligence".
Some companies that were originally positioned as big data companies have added intelligent technology to their big data solutions to make their original data models more efficient, faster, and more accurate. For example, Chengdu Shuzhilian Technology Co., Ltd., founded in 2012, is a big data service provider focusing on data governance, data analysis visualization, and data mining. The solutions launched today focus on intelligent service solutions. According to the person in charge of Datalink, the industrial manufacturing service products now launched can achieve millisecond-level online detection speeds, detection accuracy is higher than 95%, realize 20%-80% manpower release, and help customers significantly increase production capacity. Behind these capabilities is intelligent machine vision that integrates deep learning and machine vision algorithms.
In the field of smart supervision business, Shuzhilian uses the smart supervision data resource library accumulated over the years as data support, and combines it with machine learning technology to launch an intelligent system that can provide business insights and assist decision-making for regulatory authorities. product. It has been applied in scenarios such as food safety, advertising supervision, complaint information mining, and special equipment supervision.
With the popularization of intelligent applications, people have gradually realized that the core of intelligence is to let data exert value. From realizing the importance of big data to the importance of mining the value of big data, this is a process of gradual change in understanding, and it is also an inevitable process from big data application to intelligent application.
Intelligent evolution still faces a talent gap
According to the statistics of applicants for the 2021 College Entrance Examination, artificial intelligence has become a popular new major. On the one hand, it is not difficult to see that the prospects of the artificial intelligence industry are being widely recognized. On the other hand, it also reflects that the shortage of artificial intelligence talents is gradually becoming public awareness. From the observation of the industry, the gap in artificial intelligence talents mainly comes from two aspects.
First, there is a large demand for professional talents in technical research. As we all know, the innovation of the industry is inseparable from the development of underlying technology. At a time when the call for "intelligence" is getting louder, artificial intelligence urgently needs to make new progress based on deepening existing research.
It is understood that artificial intelligence theory can be divided into three stages: calculation, perception, and cognition, which technically correspond to computational intelligence, perceptual intelligence, and cognitive intelligence respectively. Hu Guoping, Senior Vice President and Dean of the Research Institute of iFlytek, pointed out at the AI ??WORLD 2018 Summit: Computational intelligence allows machines to store and calculate; perceptual intelligence allows machines to listen, speak, see and recognize; and cognitive intelligence is Solve problems that machines can understand and think about.
Alibaba Damo Academy released the "Top Ten Technology Trends in 2020" report and mentioned that artificial intelligence has reached or surpassed human levels in areas of perceptual intelligence such as "listening, speaking, and seeing", but when needed The field of cognitive intelligence involving external knowledge, logical reasoning, or domain transfer is still in its infancy.
From the perspective of industry observation, perceptual intelligence is still the mainstream technology applied in the industry at this stage. It is mostly used in fields such as speech recognition, text analysis, and intelligent image processing. It is widely used in intelligent manufacturing, smart home, smart transportation, The applications in scenarios such as smart supervision have achieved good results.
However, from the perspective of serving humans and replacing human labor, cognitive intelligence that can give machines the ability to learn and think like humans, so that they can make decisions and take actions independently, is obviously a technological tool that is more in line with the vision. Letting machines learn to think is a road full of unknowns and bumps, and it is expected that there will be a strong demand for professional and technical talents for a long time.
According to data provided by Tianyancha APP, in the past five years, the number of artificial intelligence-related companies in my country has continued to grow rapidly, with the annual growth rate remaining above 35%. Among them, more than 400,000 new related companies were added in 2020, with a growth rate of 42%, the highest in history. As of June 27 this year, there were more than 1.43 million artificial intelligence-related companies in my country whose business scope included "artificial intelligence, robotics, data processing, cloud computing, language recognition, image recognition, and natural language processing." Under this trend, the demand for comprehensive talents will continue to expand, and the talent gap in artificial intelligence urgently needs to be filled by the joint efforts of the country, enterprises, and universities. (Data Magazine/Yuan Xiaodong) Please indicate the source when reprinting