Machine Intelligence in Wu Hanqing's (Dao Ge) Eye: Redevelopment of Computer
Posted May 25, 2020 • 19 min read
Introduction: The development of human society is inseparable from technological innovation. What tests are required for the maturity of a new technology? Why is it that computing power and data are important means of production for the development of computers and AI? How to realize the intelligence of data? Through this article, Wu Hanqing(Dao Ge) will start with the development of human society, sort out and summarize the laws and context of technological innovation, and share his views on current technological development and machine intelligence.(Welfare at the end of the article:a collection of AI live broadcasts, discussing the present and future of artificial intelligence)
1. Look at the development of technology from a historical and social perspective
If you want to sort out the development of technology, you need to review human history, because human history itself is the history of the development of technology. One of Marx's definitions of human beings is that labor makes humans different from animals. What is the difference? I think it is technology that gives humans unique characteristics.
Before the agricultural revolution, humans mostly lived in the form of communities. At that time, the main means for humans to obtain food was hunting and gathering, so the efficiency was very low. About 8,000 to 10,000 years ago, there was an agricultural revolution in human history, with cropping and animal husbandry. After that, the efficiency of humans in obtaining food on a unit of land has greatly improved. Because no one is required to engage in agricultural production, the extra labor has formed the earliest social division of labor. For example, some people began to enter the handicraft manufacturing tool, which further formed the gathering and exchange of crowds. In other words, the birth of the first city in the world was driven by the technology produced by the agricultural revolution. Since then, human history has developed very slowly, and the agricultural society has experienced a long period of thousands of years. Until the modern times, about two hundred years ago, the Industrial Revolution was first born in Britain. Since then, the economy of human society has begun to show exponential growth.
Our generation who were born after World War II are particularly lucky because they have not experienced large-scale wars and enjoy the dividends brought by China's rapid development. From the beginning of the founding of the People's Republic of China, when China was poor and in vain, it immediately began the industrialization construction, and completed the national and social industrialization construction. By the time it was born after the 70s, 80s, and 90s, China has entered a consumer society and has undergone constant changes, with the emergence of computers and the Internet. Now, we are about to enter the next era, let's define it as the era of intelligence.
It is not simple to define an era. When a new technology emerges, it can have a huge impact on the economy and drive the overall transformation of the entire society, it can be called an era. From the beginning of the Industrial Revolution more than two hundred years ago to the present, it has been a huge social change, and this social change has also changed the human mind very greatly. It can be imagined that people who lived in the Ming dynasty three or four hundred years ago may have no change in life and work, and their work may not have changed much from the Han and Tang dynasties more than a thousand years ago. However, our generation is different. We directly entered the consumer society from the industrial society to the current Internet era. Therefore, even the childhood scene now looks very distant. If today it is very difficult to predict what will happen in a few decades, this is the huge change that technology has brought to society.
In the two hundred years since the beginning of the Industrial Revolution, human science and technology have continuously erupted. Returning to the essence of the industrial revolution, the essence of capital is actually technology, and technology creates new wealth. Review all the technological changes that have occurred in the past two hundred years, from steam engines, railways, trains, to steam ships, telegraphs, telephones, electrical lighting and automobiles, oil, aircraft, and then to computers, the Internet, artificial intelligence, these changes affect each other It stimulates each other and reinvents this diversified society together, and also promotes the birth of modern cities. Compared with ancient cities, modern cities have undergone fundamental changes, and these changes were brought about by the industrial revolution.
Coal, another resource of the Industrial Revolution, was very early in ancient China, but the rich people in ancient China only used coal for heating, and did not use coal as an important production resource. It is precisely because of these objective factors that the industrial revolution first occurred in Britain. The most direct reason for the industrial revolution was the stimulation of consumer demand. At that time, British consumers had a very high interest in imported colorful and cheap printed cotton fabrics from India. Therefore, the United Kingdom would spend a lot of money to buy cotton fabrics imported from India. Goods. At the time, the earliest measures adopted by the United Kingdom were trade protection measures, just as industries now contain new technologies. But even if the United Kingdom enacted laws to prohibit the import of printed cotton cloth, it could not contain the strong demand of the people, which made British producers continue to think about how to achieve technological innovation. The reason for choosing technological innovation is that the labor cost in the UK is very high, leading the capital of the textile industry to seek technical assistance to promote industrial upgrading. Replacing manpower with machines may be the only way out for the UK at the time. Therefore, the cotton textile industry has seen a lot of technological progress. For example, in 1733, John Kay invented the "flying shuttle" to increase the speed of weaving. In 1779, Cron Proton invented the "Roach" and later adapted to the steam engine It became a universal spinning machine, which greatly increased the production capacity. Later, another missionary invented the hydraulic machine, which opened a new era of mechanical weaving. In the 1920s, a junior apprentice woven 15 times more cloth by operating a machine than a skilled hand weaver.
Looking at the Industrial Revolution today, you can clearly feel the stimulating effect of demand on the birth of technology. Watt steam engine improvement is a very important task, and the cotton textile industry's demand for new technologies is the key. The reason why the industrial technological revolution took place in the United Kingdom is because it wants to reduce labor costs through technological innovation, which is also an important feature of innovation.
Back to the present, when looking at artificial intelligence, such as in a subdivision scenario in warehousing, you want to realize automatic sorting and distribution through warehousing logistics robots, and warehousing administrators often feel that it only needs to hire a few people. The cost of labor is very low. Employees are enough. From the current input-output ratio, it is not worth building an artificial intelligence system. Therefore, in the early days of the birth of new technologies, it is most necessary to find a suitable industry in order to better incubate innovation.
Overall, technology is actually productivity. In the era of the Industrial Revolution, through steam engines and spinning machines, a lot of labor costs were released, and the production capacity was increased. It can be said that demand stimulates the birth of new technologies, and the first thing that new technologies have to solve is to increase production capacity.
2 The maturity of new technologies must go through the tests of generalization, cost and safety
When it comes to computers, we must first talk about computing, which is also a huge increase in production capacity. Even artificial intelligence is still the development of computers, and neither computing nor artificial intelligence has left the von Neumann architecture, which was born in 1946.
In the history of the development of computers, Turing proposed the Turing machine, and the von Neumann architecture is a typical implementation of the Turing machine. All subsequent developments in the computing industry were implemented under the von Neumann architecture. The von Neumann architecture still realizes Leibniz's vision of simulating all things in the world through symbolic logic more than 200 years ago. Just like Leibniz s vision, computing is important because it can simulate and calculate everything that we can observe. Computing is not just basic operations such as addition, subtraction, multiplication and division. All human observations of the world can be symbolized. Therefore, all the information that can be symbolized today can be realized through symbols, which is exactly the major contribution that Boolean made.
Today, computers are an important productivity in various industries. The computer solves the problem of calculated productivity, that is, the problem of production efficiency. And calculation is so important because it can simulate everything. Therefore, many industries today are inseparable from calculations. For example, designing bridges requires very precise calculations, and because calculations make the efficiency of modern bridge design ten million times higher than in ancient times, these changes are brought by computers, so Say that computing has created a new world.
Technology has a complete life cycle. Just as during the industrial revolution, the cotton textile industry drove the steam engine technology, any technological development must initially find a corresponding demand market. Find the corresponding needs first, and then consider what problems the new technology solves. I think the fundamental of the new technology is to improve the production efficiency in the field and solve the capacity problem. After solving the fundamental problem, the next step is to solve the generalization problem. Just as the steam engine was invented with a linear drive, the earliest used to pump water in a well, and Watt s important improvement to the steam engine was the use of crank drive, which was then applied to various industries and became a universal steam engine. Two important changes Watt made to the steam engine were generalization and cost reduction.
First of all, the generalization of technology is very important. Von Neumann's biggest contribution is to realize the generalization through the transformation of the computer architecture. Later, based on this architecture, a really easy-to-use computer was created. Incubate new technology through the needs of the industry, and then serve all walks of life through the generalization of technology.
Secondly, the reduction of technical costs is also very important. Before the improvement of the steam engine, the energy consumption was concentrated on heat dissipation. When improving the steam engine, Watt designed the structure of the separation of the condenser and the cylinder, which basically solved half of the energy consumption problem and greatly reduced the cost. Looking back on the history of technology, cost is also the fundamental driving force for innovation. It determines whether technology can go to market and complete scale.
In the history of the development of computers, Von Neumann proposed two key improvements to the technology. First, the decimal was replaced by binary; second, the program and data were put into memory. At that time, the first computer in the world was made of electronic tubes, which needed to be controlled by a large number of external circuits. Von Neumann was the first to propose that circuit design and logic design should be separated, which is a huge advancement in thought.
Looking back on the development history of technology, we can find that a lot of people may have contributed together behind a technology, but Watt, Edison, von Neumann and other people who made key changes are called heroes of the technological revolution. The reason they can make a change is because they have liberated their minds, allowing them to think from another angle. Feng Neumann was originally a mathematician and did not understand computers. But when he came into contact with computers, he thought that logic design should be separated from circuit design, so he proposed von Neumann architecture. He also proposed to put data and programs directly into memory, greatly reducing the cost of computers and improving efficiency. Similarly, the transition from transistors to integrated circuits is also an important engineering improvement, which also greatly reduces costs.
There have been several key improvements in the history of the development of computers, the first was the proposal of the von Neumann architecture; the second was from the tube to the transistor; the third was from the transistor to the integrated circuit, and then only today's semiconductor industry. It can be clearly seen that in the course of several technological iterations, the most important reason for promoting technological improvement is thinking about cost reduction, which is also a typical architectural improvement idea from an engineering perspective.
Back in the era of the Internet and AI, there are several key conclusions. If you want to improve the cost of technology, the cost of raw materials becomes very critical. For computers, there are two key costs, namely the cost of computing power and the cost of data. If intelligent technology is regarded as a kind of computer technology, then there are still several typical problems that need to be solved. First, the problem of generalization, how to realize the generalization of intelligent technology; second, the problem of cost, including the cost of computing power and data, which also seriously restricts the development of new technologies.
Computers have achieved a huge increase in computing efficiency, and software is the most direct tool to help computers improve efficiency. The production materials consumed by the software are computing power and data. Computing power is essentially a transistor, and it also involves storage and networking. Integrated circuits have realized the optimization of transistors in terms of production capacity and cost, while cloud computing has realized the scale of computing power. Cloud computing still does not deviate from Von Neumann's definition of computer, in fact, it solves the problem of scale of computing power, so that the scale effect of the economy will appear. The diminishing marginal cost is also the most important reason for the establishment of the cloud computing business model, which reduces the cost of computing power. In addition to computing, the cost of data also needs to be optimized. How to effectively reduce the cost of data is a key issue to be considered in the AI era.
Looking back on the history of technological development, the first is the generation of demand, which effectively stimulated the birth of new technologies, and then the technology began to be generalized, and effectively reduce costs. But so far, the technology has not yet completed its life cycle, and it also needs to solve the safety problems and production stability problems of the technology itself. In the history of technology development, many early engineers needed to take huge risks to improve the technology, but now we only need to write software. In short, technical security issues also need to be resolved. Only after the generalization, cost, and security issues of the technology are all resolved can we truly say that the technology is mature. At the same time, after the emergence of new technologies, it will continue to iterate, which will generate new demands. For example, in the industrial revolution, the steam engine is just a sign. The steam engine has driven the demand for railways, and then trains and ships have appeared, forming a railway network And shipping network, which in turn led to changes in the times. Telegraphs also appeared during that period, thus forming a railway network for transporting substances, a ship network, and a communication network for transporting information. Today's telephone and Internet are also communication networks. These two networks have greatly promoted the global economy, and have also made global ties closer.
Between 1850 and 1910, global trade volume increased 10 times, which can reflect the industrial revolution's role in promoting world trade. The importance of the network lies in one-time construction, repeated use, and the more nodes in the network, the higher the interactivity between the nodes, the lower the cost, and the greater the economic effect. In addition, a network will bring all accessible customers into a unified market, thereby effectively expanding the scale of the industry. Compared with the railway network, the Internet has no essential changes, only changes in scale. But it can also be said that quantitative changes cause qualitative changes, and the Internet has broken the limits of physical space, making the efficiency of reaching customers on the Internet today very high. Therefore, when looking at the relationship between technology, the relationship between the railway network and the steam engine is very much like the relationship between the Internet and the computer. The emergence of the Internet has played a very large role in amplifying the economy. It mainly has four economic effects, namely, economies of scale, synergistic economic effects, bilateral market effects, and Metcalfe effect. It is precisely because of these four effects that the Internet has become a super powerful economic amplifier.
3 How to achieve large-scale breakthroughs in computing power and data
Looking back at the development of the Internet, computers and smart technologies today, I think the first thing that has really been verified is search. The development of the Internet has spurred the development of search technology. Search is the first software to achieve computing power and data scale, and it still meets the von Neumann architecture. In fact, when search was first proposed to solve the problem of computing efficiency of massive data, it was first invented by Yahoo. Later, Google made improvements through the PageRank algorithm to create a huge market with a better experience. It is precisely because of the huge market demand found in the field of Internet advertising that the continuous flow of technology has been invested in the field of search. It is precisely because of the precipitation of search technology that the problems of computing power and data scale can be better solved, and cloud computing and big data have been formed.
Traditional thinking cannot think about the valuation of new technologies. For innovative technologies, it is difficult to form a correct valuation for it, because the technology iteration speed is very fast. For example, when the steam engine appeared, railways and ships soon appeared. Therefore, when evaluating the steam engine, it is difficult to imagine how big the market will be and how much change will be made to the world and society. This is the power of innovation. Therefore, only technological innovation is creating new wealth, because it increases production capacity, and then continuously expands the scale through the investment of capital. Therefore, when studying the changes and iterations of technology, it is difficult to imagine what new demand will be created after a new technology comes out, so it is difficult to evaluate the future potential of a new technology through a set of mature methods.
Today, if you want to have better competitiveness in the competition of enterprises, you need to think about each new technology from the perspective of resources. For example, in the industrial age, there was a big discovery of natural resources. Britain first used wood as the main fuel for production and life. Later, it began to use coal because of the reduction in wood. However, it was not until the invention of the steam engine that coal was mined in large quantities. Another important reason for the birth of the Industrial Revolution in the United Kingdom is that the United Kingdom has more coal and is more convenient to mine. It is closer to the coastline and easier to transport. In ancient China, coal mines were mainly inland, and the most economically developed areas were in the Yangtze River Delta. Without railways and ships, the cost of transporting coal to coastal areas was very high, so coal did not become important for the economic development of ancient China. material. When the steam engine appeared, coal became an important strategic resource. The same thing happened after the advent of oil. Cleveland, the United States, developed from a small town into an industrialized city because of nearby oil fields. It was here that Rockefeller got its first pot of gold. The same is true for power development. In 1886, the United States decided to build a hydropower station on Niagara Falls to deliver 100,000 horsepower to Buffalo, 26 miles away. Before the invention of electric power, waterfalls can only be regarded as spectacular natural landscapes, and after the invention of electric power systems, you will find that the water drop of waterfalls is a very important natural resource, and whoever owns such resources will have a huge wealth. The monopoly or seizure of natural resources can put enterprises or countries in a competitive position. Therefore, it is necessary to look at the development of new technologies from the perspective of natural resources. When a new technology appears, innovation actually needs to focus on market size. For the change of the times, the scale of innovation is the most critical and essential thing.
Back to the development of computers, the first electron tubes will also emit heat and have huge energy consumption problems. Later, when the transistor was invented, silicon became a new natural resource, and this material is everywhere on the earth, making new transistors cheaper than repairing tubes, which led to the restructuring of the business model. This shows that natural resources determine many things.
For the development of computers and AI, computing power and data are important means of production. Computing power involves transistors, storage networks, etc., and data needs to be rethought as a new natural resource. In the information age, the development of the computer is the emergence of a variety of databases, replacing many of the manual filling work through the computer. In the era of intelligence, some low-threshold, low-complexity things can be done automatically by computers. Looking back, the scale of computing power and data utilization by computers in the information age is relatively small. The computing power and data used in the PC era may only come from themselves, but in the smart era of software operations, cloud computing may be used. Massive data and computing power. Therefore, today's new software can solve different problems, and its improvement in efficiency will be countless times that of the information age. Cloud computing has effectively completed the scale of computing power. The scale of data used to be expected to be completed by big data, but currently big data has not been fully applied, and various industries are still processing small data. Today we talk about what kind of software should be designed for machine intelligence. In fact, we should look at how much computing power and data this intelligent software consumes, and judge from the scale. If an intelligent software does not consume a lot of computing power and data, it is difficult to imagine that it can increase very high production capacity.
To achieve the scale of data, the most important thing is IoT. Today, through a large number of IoT devices, massive data will be continuously collected, and the data scale is much larger than the data scale generated in the era of manual form filling. The data collected by manual filling is still considered to be the data collected in the information age, and the data collected by IoT devices is the basis for meeting the data scale required in the smart era. For data to become a natural resource, scale is important.
For the cost of data, the high cost of AI is because the cost of data is very high. The cost of computing power has effectively achieved economies of scale through transistors, integrated circuits, and cloud computing. It can gradually converge in the implementation path and comply with Moore's Law. The cost of data is still very high, which is mainly due to several reasons:First, for many AI algorithms, the cost of data marking is very high, often requiring manual participation, so the entire academic and industrial circles are thinking about how to make the machine automatic Labeling, thereby effectively reducing data costs. Second, the cost of IoT devices and full-link data collection is too high. IoT devices do not produce data, but collect data. The development of 3G/4G/5G has reduced the cost of device data collection links. After data collection, it also needs to consider how to store and process, it needs to comprehensively consider the cost, and also needs to interact with the upper application software, so manual data governance also wastes a lot of time. As long as manual participation is required, the efficiency must not be high and the cost will rise. Therefore, academia and industry are also thinking about how to let machines automatically perform data governance to solve data quality problems. When all of these are completed, data can be used as a natural resource that can be effectively used in the intelligent age.
Four intelligent data
With large-scale computing power and data, how to improve the productivity of various industries? Machine intelligence technology is actually a bit like the steam engine that appeared in the industrial revolution. The application of various industries is similar to the textile machine of the British cotton textile industry. Therefore, there is no way to simply say how intelligent technology can change the world, but to manufacture "textile machines" in all walks of life to increase production capacity and reduce costs, thereby stimulating industrial progress and serving the entire society. Therefore, it is necessary to observe which industry can promote the development of its own technology and realize digitalization and intelligence.
The intelligence of data is to solve the rational allocation of resources through refined management or governance. This is the most common contribution that computing can make to all walks of life. When this problem is resolved, the production capacity of various industries will increase dramatically, and the cost will also drop significantly.
Returning to the economic and social level, after the emergence of steam engines, railways, and ships, agricultural mechanization revolution took place, a large number of farmers began to lose their jobs, a large number of people flocked to cities, and the working class emerged. Because the factory invests in large-scale machinery and equipment, a large number of workers are required to gather to work, so industrial division of labor has emerged, further forming an industrialized city. So the earliest industrialized city was Manchester, England. At that time, 500,000 workers used machines to manufacture cotton textile industry. With the development of the power and oil revolutions, in the two or three decades of the early 20th century, the number of cities with a population of over one million rose from six to two or thirty. This is the large-scale urban population brought about by the industrial revolution. Gather.
In the era of consumption, industrial agglomeration has further occurred, leading to the earliest batch of enterprises that began the industrial division of labor to accumulate a lot of wealth, so class divisions began to appear, and there were workers, white-collar workers, and the middle class. Because many important production activities are not completed in the front line, but in the office, and the accumulation of knowledge and experience of the middle class has become very important, this group of people has become very rich, and therefore has driven social demand , Especially the demand of the tertiary industry. Around 1990, the industrial structure of society has undergone tremendous changes. A large number of people have moved from industry to service, and the urban structure has shifted from industrial to consumer.
Today, Internet companies are driven by technology, and many companies do not strictly enforce punch card punching, because we have brought people back to people instead of machines, which is actually the problem that artificial intelligence has to solve. Because the system for people to gather punch cards to work is a set of management systems invented in the factory based on assembly lines after the Industrial Revolution. Today, if technology is capital itself, technology creates new wealth, and innovation is the most important driving force for wealth development, then we should provide a more relaxed environment for technological innovation.
With the advancement of machine intelligence, the social division of labor unions will be further refined, and cities will gather more people. It is estimated that by 2050, 70%of the world's population will live in cities. The development of cities needs to solve problems such as housing, employment, environment, and health. Therefore, after the emergence of intelligent technology, we hope to rethink and solve the ills brought about by the industrial revolution, so that the entire society and humanity can achieve sustainable development.
Whether it is computing power or data, through refined calculation, you can effectively reduce losses and achieve a reasonable allocation and utilization of resources. Machine intelligence will have a huge impact on the future economy and society, allowing more people to return to more creative work, making people more human-like, making work more open and free, and changing the face of life. The development of technology still conforms to the aforementioned rule, and strong demand is needed to stimulate the birth of new technology. It is necessary to solve the problems of versatility, cost and safety of technology. Finally, technology creates new demand and establishes a positive relationship between technology and demand Feedback, and then form a new social division of labor, in order to change the face of society, cities and life, and promote new culture and values.
Today, whether it is AI, IoT, or 5G technology, we should still think from the perspective of demand, cost, and economic impact, and make effective strategic decisions. We should also think from the perspective of natural resources to find new competitive advantages for enterprises.
This article was collated by DataFunTalk Wu Xuesong and Alibaba Cloud Developer Community Volunteer Jia Zijia.
Welfare is coming | AI Live Collection
Five Ali Technology Masters
Talking about the present and future of artificial intelligence
AI continues to be hot and enters an explosive period. This collection gathers Ali AI technology experts and product experts, from the development trend of the entire AI field, related technologies to the practical application of AI, to share relevant views on the development of AI, and to display the latest technological achievements of Ali AI. Full of dry goods!
Original link: click here