Research Report on Artificial Intelligence Industry

Research Report on Artificial Intelligence Industry

Overview

AI technology is an absolutely unavoidable technology for investors who are concerned about high-tech. It is self-evident that the machine replacement and automation on the supply side have the same trend as the consumer upgrading on the demand side. We look at AI from two corresponding perspectives: Artificial Intelligence (AI) and Augmented Intelligence (AI). The former aims at replacing human labor with machines, while the latter aims at assisting human work and life with machines. We also divide AI research into two parts: Neural Network (ANN) and Robo Network (Robo). The former substitutes or assists the mental work that only human beings can do at present, while the latter substitutes or assists the physical work of human beings with machines. As a result, AI technology mainstream topics or subdivisions can be divided into four quadrants in Figure 1:

Invest in Originality with Love: Research Report of Artificial Intelligence Industry

Figure 1: Research from Love

Among them, the topics in the upper left quadrant belong to the category of machine learning, while the upper right quadrant involves the combination of machine learning with other algorithms such as big data and distributed books, while the lower left quadrant and the lower right quadrant involve the combination of machine learning with more human-computer interaction, materials, energy and even biotechnology. It can be seen that robots are the most extensive concept of technology extension.

Invest in Originality with Love: Research Report of Artificial Intelligence Industry

Figure 2, from Erie Report on China's AI Application Market in 2015 [1]

Figure 2 summarizes the 60-year development of AI technology, including two ups and downs, and the third golden period brought about by the success of Deep Learning since 2013.

The social reality of the gradual disappearance of the domestic population dividend, the trend-oriented policy of import substitution and machine replacement, and the economic and industrial requirements with the prominent features of precision, flexibility and globalization of production are all the reasons for the new round of attention and pursuit of AI concept by capital at present.

Invest in Originality with Love: Research Report of Artificial Intelligence Industry

Figure 3, from the IDC report The DIGITAL UNIVERSE IN 2020: Big Data, Bigger Digital Shadows, and Biggest Growth in the Far East [2]

Technically, in addition to breakthroughs in algorithms, the growth of data resources and computing power also plays a key role in the development of AI. IDC predicts that by the end of 2020, the total global data volume will increase 50 times on the basis of early 2010 [2]. Among them, the development of the Internet of Things, especially the popularization and growth of sensors, may be the main pillar.

Invest in Originality with Love: Research Report of Artificial Intelligence Industry

Figure 4, from The AI Revolution: The Road to Superintelligence [3]

Figure 4 shows the historical growth of computing power in the 20th and 21st centuries and the expected growth according to Moore's Law. At the exponential growth rate, $1,000 will be able to buy computing power comparable to that of the human brain by 2025. The further development of cloud computing, GPU and NPU may be the main way of implementation.

Many investment and consulting organizations look at the whole AI industry from three levels: infrastructure, technology and application. Infrastructure layer mainly provides data resources and computing power. In this layer, giant monopoly is the most serious, because network effect and scale economic advantage play the most obvious role here. Technology layer takes machine learning algorithm as the center and provides universal technical support. The market concentration of this layer is also high, because AI has the hard support of talents, technology and resources. The application layer mainly provides a variety of industry solutions. There are relatively many opportunities for innovation in this layer. The entrepreneurs who cross the threshold have broad space to seek differentiation.

AI technology and application start-ups generally face the "chicken and egg" problem of algorithms and data. On the one hand, using the advantage of big data can train and optimize more intelligent solutions; on the other hand, only by constantly improving specific algorithms, can they maintain the advantages of active users and data resources. More importantly, if the data resources of Companies in different subdivisions are utilized comprehensively, the value generated may be the square of relatively isolated data value. Therefore, strategic cooperation, strategic investment, mergers and acquisitions may be more obvious norms and themes in AI field.

We advocate the integration of technology, data, talent, market and other factors through joint investment and capital as a link to create a competitive advantage for independent companies. On the basis of excellent team, leading technology, clear market positioning and credible business model, we look forward to enterprises with large ecosystem as a backing or strong public awareness.

Technical Paper

Following is the basic principle of neural network and machine learning, which is the core of AI technology. Aiyi Investment is especially willing to communicate with investment managers who pay more attention to AI and strategic leaders of enterprises in various industries to seek opportunities for cooperation. Startup CTOs don't need to review the basics.

1. Neurons and Feedforward Neural Networks

Artificial neural network (ANN) is a computational model designed by human beings to imitate the working mode of biological neural network. Neuron is the basic unit of neural network, also known as Node. It receives input from outside or other nodes and calculates output through an Activation Function. Each input corresponds to weight, i.e. the relative importance of each input received by this node; Bias can be understood. For a special input [4].

Love is original: artificial

Please read the Chinese version for details.