Li Xunlei is the chief economist of Zhongtai Securities Co., Ltd. and the vice chairman of the China Chief Economist Forum
Since the first industrial revolution, science and technology have gradually become an important factor of production for global economic growth, and each major technological revolution will promote the long-term growth of human social economy by leaps and bounds. The criterion for whether new technologies can trigger a new round of scientific and technological revolution lies in whether they can be widely used in production and life, and liberate and develop productive forces.
According to the law, there is a so-called "Solow paradox" in the process of improving productivity through technology, that is, there is a significant time lag in the improvement of productivity by technology. Although AI-related technologies have been developed for a long time, their impact on production methods has not been separated from the framework of the Internet. With the maturity of basic general technologies such as computing power and natural language processing, the rapid application of ChatGPT based on natural language and the rapid expansion of AIGC application scenarios provide an achievable path for AI technology to accelerate the breakthrough of the "Solow paradox". The wave of AI has arrived, this round of scientific and technological revolution is destined to bring "subversive" changes to all walks of life, and new technology leaders are destined to be born in the "AI+" sector of the capital market in the future.
The dividends of the third industrial revolution are on the verge of exhaustion, and global economic growth urgently needs to inject new momentum
Looking at the history of global economic growth for more than 2,000 years, before the 19th century, global economic growth was very slow. According to the famous economic historian Angus Madison, from the first century to 1820, the average annual growth rate of the global economy was only 0.1%, and per capita income increased by only about 40% in the past 1800 years.
Until the occurrence of the British Industrial Revolution more than 200 years ago, this situation was completely changed, and the production methods that were mainly dependent on human and animal power changed, and the global economic growth began to accelerate. Since then, every major technological revolution has promoted the rapid long-cycle growth of human social economy, and science and technology have become the most important driving force for world economic growth.
Many people think that we have been at the peak of the technological explosion because as the penetration of the Internet in daily life continues to deepen, we have the illusion that we are still in the era of the "big bang of technology". There are two main reasons for this illusion: first, China has absorbed almost all of the West's 300 years of technological achievements in just 30 years; Second, the development of information technology is actually only "improvement of leveling efficiency" rather than "breakthrough in higher dimensions", that is, although computer memory and operating efficiency continue to improve, the essence and working principle of operation are still in place, and the technology that truly improves the quality of total factor productivity has not appeared.
Even the scientific and technological dividend of the 21st century, represented by "Internet +", is on the verge of exhaustion. In 2021, the global Internet penetration rate reached 65.6%, of which the penetration rate in China reached 65.2% and the penetration rate in the United States reached 90%. The extremely high mobile Internet penetration rate of the world's major economies indicates that emerging Internet platforms have entered a red sea era in which it is difficult to obtain "incremental" and needs to rely on "stock", in which case the marginal cost of acquiring new users or improving user usage time is extremely high.
At the same time, with the evolution of scientific and technological innovation and industrial transformation cycle, the Internet has gradually become a traditional industry. The Internet has not only realized the transition from desktop Internet to mobile Internet, but also realized the coverage from information Internet, consumer Internet to industrial Internet, and application scenarios have penetrated into all walks of life, and it is difficult to find more expansion points.
The improvement of labor productivity ushered in important opportunities, and artificial intelligence may bring a new round of scientific and technological revolution
The symbol of the outbreak of the scientific and technological revolution is that a new generation of scientific and technological achievements has begun to be widely used in production and life, thereby liberating and developing productive forces and improving labor productivity. The technological breakthroughs that have occurred globally in recent years have not essentially liberated and developed productive forces, especially for large economies that can lead the global economy.
The launch of ChatGPT, a new AI top-notch at the end of November last year, has once again sparked people's reverie for a new round of technological revolution. Only 2 months after the launch of ChatGPT, the number of users has exceeded 100 million, at that time it took 75 years and 16 years for phones and mobile phones to accumulate 100 million users worldwide, and even the last fastest program TikTok took 9 months.
ChatGPT has a super-large model and network-wide data training, achieving a phenomenal breakthrough in generative AI. Unlike the popular metaverse in the previous two years, ChatGPT has a much wider range of application scenarios than the metaverse, reflecting the emergence of brain-liberating "smart" production tools. Just like the Internet 20 years ago and smartphones 10 years ago, it is likely to trigger a new round of technology and industry changes.
The use of the term "scientific and technological revolution" in the media is very widespread, in fact, everything that does not significantly bring about a total factor productivity increase is actually only a "pseudo-revolution". Looking back at history, the industrial revolution is a period of technological innovation, and the three industrial revolutions have all promoted the improvement of labor productivity without exception, and determined the growth and prosperity of a country and even the world economy in the future:
The first industrial revolution from 1760 to 1840, marked by the invention and widespread application of the steam engine, the production tools underwent a qualitative change from manual form to machine form, human society began to move towards the era of mechanization, and the average labor productivity of British workers increased by 20 times between 1770 and 1840.
The second industrial revolution from the end of the 19th century to the beginning of the 20th century, marked by the emergence of internal combustion engines and electrical machinery, made large-scale production the dominant production mode, greatly reduced industrial production costs and improved production efficiency, and in the 30 years at the end of the 19th century, the world's total industrial output increased by more than 2 times.
Since the 60s of the 20th century, with the emergence and wide application of semiconductor technology, mainframe computers, personal computers and the Internet, automated machinery and equipment have not only replaced a considerable proportion of physical labor, but also replaced a certain degree of mental work, social productivity and human civilization have reached an unprecedented new height, and human beings have entered the information age.
Different from other technological breakthroughs, this round of artificial intelligence represented by ChatGPT is very likely to achieve intelligent replacement and function multiplication of production factors such as labor and capital, and promote the improvement of total factor productivity. The steam engine promoted the first scientific and technological revolution because it increased labor productivity while liberating a large number of labor from low-level agricultural labor. Electricity accelerates the transfer of labor from the primary industry to the secondary industry and stimulates the development of the tertiary industry. In the era of information technology, a large number of labor forces have poured into the tertiary industry, forming a pattern in which the global tertiary industry accounts for 55% of GDP.
In the future, with the widespread application of artificial intelligence, more and more fixed, tedious and standardized jobs will be replaced by artificial intelligence, which can not only alleviate the problem of labor shortage in the era of aging population, but also help workers focus on more advantageous and creative work, thereby greatly improving labor productivity. At the same time, data will become the leading factor of production for economic growth and the basis of a new generation of artificial intelligence, and by deeply mining information from massive data and converting it into knowledge capital, it can effectively multiply the functions of production factors, resulting in more significant efficiency gains in productivity.
For example, a Goldman Sachs report on AI argues that with AIGC, workflows will be greatly simplified and productivity will be boosted, and productivity is expected to increase by more than 1.5% per year within a decade of generative AI development. We have reason to believe that artificial intelligence will set off a new round of scientific and technological revolution in the future, bringing about the restructuring of production methods and industrial structure, and improving the productivity of today's society. After the maturity of artificial intelligence technology, it is destined to form an important supporting force for the improvement of labor productivity in the whole society and even the world.
The whole AI industry chain is gradually maturing,
Artificial intelligence is expected to accelerate the breakthrough of the "Solo paradox"
Artificial intelligence is not a new technology, but its effect on total factor productivity is not obvious. One of the important reasons for this is that there is a "time lag" in the diffusion, transmission and promotion of advanced artificial intelligence in the overall economy. Looking back at the last wave of artificial intelligence, probably after the 2017 AlphaGo game, this popularity has not been maintained, mainly because there is no large-scale application landing, which has not directly promoted the progress of productivity.
However, this does not mean that artificial intelligence cannot trigger the next round of scientific and technological revolution. In fact, even during the Third Industrial Revolution, there was a significant time lag in the productivity gains of information technology, the so-called "Solow paradox." At that time, the growth rate of labor productivity in the United States fell from an average of 3% in 1948-1973 to about 1.5% in 1984-1990, and it was not until the 90s that the role of information technology in productivity improvement became apparent.
In the same way, it will take a while for artificial intelligence to play a role in the improvement of total factor productivity, which may not be able to see the obvious role until three conditions are gradually met: first, it is necessary to have corresponding new infrastructure such as algorithms, computing power, and data to promote the full commercial application of artificial intelligence technology; The second is to strengthen complementary innovation, use the self-learning ability of machine learning systems to stimulate complementary innovation, and realize the benign interaction between artificial intelligence technology research and development and industrial development; The third is to increase supporting investment and realize the reengineering of enterprise production mode, organizational structure and business process.
The new round of artificial intelligence boom triggered by AIGC represented by ChatGPT allows us to see the possibility of the above conditions being gradually met: on the one hand, in recent years, the level of computing power and algorithms around the world has been continuously improved, as the foundation of the AI era, countries have paid great attention to the market-oriented development of data elements, and strived to mine, cultivate and release the value of data, from the construction of the national integrated government affairs big data system from east to west computing, to the promulgation of the "Data Twenty" and the "Digital China Construction Overall Layout Plan" successively. The mainland has also been actively leading the development of data elements. And with the continuous iteration of AIGC, we can see the hope of widespread application of artificial intelligence in the fields of pension, education, medical care, and content creation.
On the other hand, since 2006, the practical process of deep learning has created conditions for the benign interaction between artificial intelligence technology research and development and industrial development. Compared with AlphaGo, the emergence of ChatGPT means that humans can not only use neural networks to deal with closed problems, but also deal with open problems through continuous optimization, pre-training, adversarial training, and reinforcement training. Recent research papers show that GPT-4 has some germination of self-reflection and error correction ability, which makes it possible to establish a growth flywheel of data, learning and intelligence in interaction with humans, so as to achieve benign interaction with industrial development in the process of industrialization.
In addition, due to the obvious correlation between AI investment and business impact, AI investment will bring obvious benefits to enterprises, with the continuous development and maturity of AIGC technology, a large number of companies will deploy AI, enterprise production mode, organizational structure and business processes will change dramatically. Since the AIGC concept became popular at the end of last year, there have been Microsoft, Google, Baidu, Ali, Huawei and other enterprises to lay out the field of artificial intelligence, the future AI technology will subvert all walks of life, such as artificial intelligence may be packaged into "solutions" to strengthen the existing moat of SaaS companies, office tools will become the most widely used AIGC, etc., will largely change the business model and management of enterprises.
Japan's experience under technological innovation in the 1970s: China's manufacturing industry may obviously benefit from "AI+" empowerment
From 1955 to 1970, through the technological development strategy of "introduction, digestion, absorption and re-innovation", Japan mastered almost all the advanced technologies invented and applied by industrially developed countries in the past half century, basically eliminated the gap of 20 or 30 years behind Europe and the United States, followed the example of Europe and the United States and established a set of heavy industrial structure, and the gross national product rose to the third place in the world.
As the technological level of Japan and Europe and the United States and other countries is gradually approaching, the previous development model of "introduction, digestion, absorption and re-innovation" in Japan seems inappropriate. At the same time, although the heavy industrial structure has supported the rapid development of the Japanese economy, the inherent limitations and problems of this industrial structure have become increasingly prominent, and it has also forced Japanese monopoly capital to carry out industrial adjustment.
With the in-depth development of the third scientific and technological revolution in the 80s of the 20th century towards electronic technology, new material technology and life science technology, Japan abandoned the scientific and technological development model of take-ism, put forward the development strategy of "science and technology to create the country", and the rapid development of the electronics industry represented by information technology and new material technology led Japan to maintain a high economic growth rate in the economic environment of "three-phase superposition".
Compared with the 60s, the absolute and relative contribution of Japanese capital to economic growth fell sharply in the 70s, but more than half of the contribution to economic growth was due to technological progress. During this period, Japan's labor productivity also rose rapidly, benefiting from continuous technological and equipment transformations: according to the Modern Japanese Economy, Japan's labor productivity grew by an average of 6.8% between 1973 and 1980, while the United States and the Federal Republic of Germany grew by only 1.7% and 4.8% respectively during the same period.
After the 2008 financial crisis, China also faced the economic problems of "three-phase superposition" of growth rate shifting, structural adjustment pain, and digestion of stimulus policies. Learning from Japan's experience, in order to maintain medium-high economic growth and promote the economy to the middle and high end, the mainland must seize the development opportunities of the new round of scientific and technological revolution and achieve a high level of self-reliance and self-improvement in the fourth wave of scientific and technological revolution.
But unlike Japan in the 70s, China today faces a more severe geopolitical environment. Since China's accession to the WTO in 2001, China's economy has caught up with the United States: in 1991, the United States accounted for 25.9% of the world's GDP, and Japan accounted for 15.1%, when China's GDP was only 1.6% of global GDP. By 2021, China's share of GDP had reached 18.5%, while the share of GDP in the United States and Japan fell back to 23.9% and 5.1% respectively. In addition, China is also challenging the hegemony of the United States in science and technology, national defense, education and other aspects.
Behind each epoch-making development of science and technology is the fierce competition between major powers, the fundamental reason is that the urgency of military needs regardless of cost is greater than the urgency of civilian needs: under the pressure of war, the state will increase financial and material investment to support scientific and technological research and development, and apply advanced technology to the military field. After the fierce confrontation between major powers, these advanced technologies will be applied to all walks of life in the economic system to realize the industrialization of the scientific and technological revolution.
In contrast, there is still a big difference between China and the United States in ideology and military strength from Japan in those years, so the United States has more and more powerful means to suppress China than it did to suppress Japan. If China only competes with the United States in the existing technological framework, it will only be constantly "stuck in the neck", and only by fully grasping the wave of this scientific and technological revolution can it be expected to win from the global competition.
Nowadays, the outline of the fourth industrial revolution characterized by intelligence is becoming clearer, and the world's unprecedented changes in a century and the historic inflection point of a new round of industrial revolution, whether it is national, local or enterprise, will spare no effort to deploy investment, which also means that artificial intelligence will have historic development opportunities in the mainland in the future.
The singularity is coming: pay attention to investment opportunities in the new round of technological revolution
Looking back at the occurrence of previous scientific and technological revolutions, it will inevitably bring a lot of financing. In essence, only financial support can increase the R&D advantages of core enterprises and further expand the maturity and application of technology. In the context of the new round of scientific and technological revolution, emerging technology enterprises will generate a large amount of financing needs. The new pattern of AI development may open a new wave of investment golden period. In particular, after the implementation of the comprehensive registration system, it is necessary to pay more attention to the investment opportunities of related financial technology companies and science and technology companies in the field of AI.
In the short term, the current market concentration of the AI sector is high, and the valuation of some companies benefiting from downstream applications under the thematic investment market is too high, and investors need to avoid the risk of overheating; However, in the medium and long term, if artificial intelligence compares to bring a round of scientific and technological revolution, it is destined to give birth to new leaders in global technology, and leading companies with core R&D advantages in the field of AI should have a rational and objective long-term investment philosophy.
Computing power as a key factor restricting the development of AI, at present, the AI industry leaders mainly use NVIDIA's GPU chips for development and training, and the domestic high-performance GPU research and development capabilities are limited, with high domestic substitution urgency. At present, the chip manufacturing process has been upgraded to 5nm, and the cost performance of the process has gradually decreased, in this case, "advanced packaging" may become the key technology of "curve overtaking" in the field of mainland computing power.
AI algorithms include basic algorithms and application algorithms, almost all basic algorithms are provided by Stanford, Harvard, OpenAI and other American institutions, and the application algorithms are almost completely open source, and the United States has absolute control over the open source community. In other words, the current dependence of mainland AI algorithms on foreign countries is very large, and it is recommended to pay attention to manufacturers with core algorithm breakthroughs and commercial landing of AI algorithms.
At the same time, massive data can create a good development environment and data technology foundation for artificial intelligence. With the establishment of data as a national basic strategic resource, data elements to promote the development of the digital economy will enter the landing stage, the future state-led data trading system will gradually improve, data elements related to collection, confirmation, storage, processing, pricing and circulation six links will benefit a large number of companies, and is expected to form a plate effect.
In addition, in terms of AI downstream applications, with the continuous expansion of AI+ application scenarios, the demand market for application scenarios continues to expand, and software leaders with good business models may also benefit significantly.
Risk warning: the commercialization of AI technology is not as expected; Technology iteration is less than expected; Relevant policies are not advancing as expected.
——————