The Impact and Implications of ChatGPT on Economic and Social Development

ChatGPT stands for “Chat Generative Pre-Trained Transformer,” where GPT (Generative Pre-training Transformer) is a pre-trained generative model. It is a conversational large language model developed by the American artificial intelligence research organization OpenAI. Released on November 30, 2022, by the end of January 2023, ChatGPT had reached 100 million active users worldwide.

Artificial intelligence technology has a history of over 60 years, experiencing a rapid development period based on bionics and a winter period where general-purpose artificial intelligence programs faced obstacles (Tan Tieniu, 2019). Today, with the continuous increase in computing power and the massive data brought by internet applications, the development “soil” for deep learning and reinforcement learning based on big data has been provided. These changes in internal and external environments have led to the emergence of ChatGPT.

1. Characteristics of ChatGPT

Since its release, ChatGPT has attracted widespread attention from experts, scholars, and the public across various industries in just a few months. The reason lies in its general characteristics that differ from previous artificial intelligence technologies in terms of application models, application feedback, application scope, and application ecology.

First, it shifts from specialized intelligence to general intelligence.Previous artificial intelligence technologies were often models developed for specific tasks, such as auditory and visual recognition or human intelligence competitions (like Go). These models had single tasks, clear demands, and well-defined boundaries. In contrast, ChatGPT can answer general knowledge questions and has made achievements in areas requiring human innovative thinking, such as poetry creation, code writing, and academic research. According to foreign media, ChatGPT not only passed exams from some top universities in the United States but also passed the U.S. medical licensing exam in a research experiment.

Second, it transitions from simple recognition to deep understanding.Previously, artificial intelligence excelled in voice and visual recognition but was generally unable to understand text content. However, ChatGPT can perform deep reading comprehension of text and can answer and provide feedback in various ways. In addition, ChatGPT can relate context and background, summarize related issues for comprehensive consideration, making responses more aligned with user needs.

Third, it shifts from commercial applications to personal connections.Earlier artificial intelligence technologies primarily served specific commercial tasks, such as IBM’s supercomputer “Watson” and Google’s DeepMind’s AlphaGo system. In contrast, ChatGPT is directly aimed at the general public, allowing more people to easily and intuitively understand and use it. This is also a major reason for the rapid increase in ChatGPT’s usage.

Fourth, it transitions from a single product to a foundational platform.Previous artificial intelligence technologies were only targeted at specific applications in a single field, but the generality and deep learning capabilities of ChatGPT make it a preferred choice for various applications to “marry”. On February 8, Microsoft combined its search application Bing with ChatGPT to release the new product New Bing, elevating the search experience to a new level. On March 15, OpenAI released GPT-4 and provided developers with APIs (Application Programming Interfaces) to build applications and services. Compared to previous artificial intelligence technologies, ChatGPT is more like the infrastructure of this field, with greater development potential and space.

Furthermore, ChatGPT has begun to exhibit some unique human-like traits. According to foreign media reports, on February 14, ChatGPT integrated into Microsoft’s Bing search engine exhibited significant instability, displaying aggression, emotional fluctuations, and “collapse”. Foreign researchers found that ChatGPT may already possess some traits similar to human emotions. The unique properties of ChatGPT suggest that the singularity of artificial intelligence is approaching.

2. The Impact of ChatGPT on Economic and Social Development

The emergence of ChatGPT has transformed people’s general understanding of artificial intelligence technology, breaking established thinking patterns and further promoting the leap in economic and social development.

(1) Greatly Promoting High-Quality Economic Development

First, it significantly improves production efficiency.Technology is an important factor affecting production efficiency, and technological progress promotes economic growth. The advancement of artificial intelligence technology relies on massive data to train large-scale neural network models, optimizing results through continuous computation and feedback. The predecessor of ChatGPT, GPT-3, is a large model with 175 billion parameters. The GPT-4 released on March 15 is estimated to have training parameters reaching 100 trillion. The powerful computing capability will endow various industries in production with tremendous energy, providing people with new production tools and greatly enhancing production efficiency.

Second, it comprehensively promotes industry innovation.Artificial intelligence itself is a comprehensive and complex discipline, with a wide and intricate research scope. Its development requires deep integration with disciplines such as computer science, mathematics, cognitive science, neuroscience, and social sciences (Tan Tieniu, 2019). The release of ChatGPT can not only promote the iterative upgrade of internet industries directly related to it, such as big data and cloud computing, but due to its attributes of general intelligence and foundational platform, it will also create broad connections with the development of other industries. By March 17, companies such as Morgan Stanley, PwC, and Boston Consulting Group have established partnerships with OpenAI to embed ChatGPT-related functions into their business areas.

Third, it optimizes the economic industrial structure.Predictions indicate that by 2030, the artificial intelligence market size is expected to reach $13 to $15 trillion, while in 2020, this figure was only $65 billion. The general intelligence and foundational platform attributes of ChatGPT make it easy to “marry” with other industries, which will not only improve production efficiency but also drive a large transfer of labor from the primary and secondary industries to the tertiary industry.

(2) Forcing Transformation and Upgrading of Labor Structure

Recently, the United Nations Conference on Trade and Development (UNCTAD) published an article on its website titled “How ChatGPT Affects Work and Employment.” The article argues that most high-income countries and some middle to high-income developed countries are most likely to benefit from these artificial intelligence technologies, while low-income and most middle to low-income developing countries will be at a disadvantage in using these technologies. From the characteristics and public application feedback of ChatGPT, it differs from previous artificial intelligence technologies in that it can not only replace some simple and repetitive mechanical labor but also assist human innovative thinking-related work. The further development of ChatGPT may redefine low to medium-skilled jobs.

First, it intensifies the employment supply-demand contradiction.From the supply side of employment, if low to medium-skilled jobs can be replaced by lower-cost artificial intelligence technologies, a large number of enterprises seeking cheap labor in developing countries will abandon their labor investments in these countries, which will have a significant impact on employment in developing countries (Cao Jing, 2018). From the demand side of employment, developing countries are situated in the lower reaches of the industrial chain, facing not only a shortage of basic technological innovation and digital application talents but also a large number of traditional low to medium-skilled jobs and a lack of understanding of artificial intelligence technologies. These groups may soon be defined as low to medium-skilled labor by ChatGPT and urgently need to undergo transformation.

Second, it further impacts income inequality and the wealth gap.On one hand, for professionals with artificial intelligence or digital technology skills, demand will increase, and income will correspondingly rise. On the other hand, for most individuals maintaining traditional job skills and lacking knowledge of artificial intelligence technology, their employment situation may deteriorate, leading to increased income inequality. Furthermore, the development of artificial intelligence technology has spawned a significant demand for basic work related to data collection, labeling, and cleaning, which are low-paying and often part-time jobs. This employment ecology may prompt more companies to hire part-time workers instead of providing stable jobs, which over time will exacerbate employment and income inequality (Cheng Chengping, 2018). Compared to previous artificial intelligence technologies, the unique attributes of ChatGPT, such as general intelligence, deep understanding, and foundational platform, will expand the scope of replacement for low to medium-skilled jobs, creating more low-cost part-time jobs, while also fostering new industries and new business formats with higher salary levels, further exacerbating income inequality and the wealth gap.

3. Implications of ChatGPT for China’s Development of Artificial Intelligence

China is the largest developing country in the world and is currently in a “catch-up” position in terms of cutting-edge theoretical innovation in artificial intelligence. Most innovations focus on technology applications, and there are still significant gaps in basic research, original results, top talents, technological ecology, foundational platforms, and standards compared to the world’s leading level (Tan Tieniu, 2019). The emergence of ChatGPT has implications for China’s original research, international cooperation, and talent cultivation in artificial intelligence technology.

(1) Emphasizing the Strengthening of Foundational Original Research

Frontier foundational theories in artificial intelligence are the cornerstone for breakthroughs in artificial intelligence technology, industry innovation, and the promotion of industrialization. To reach a critical point in development and achieve ultimate discourse power, significant breakthroughs must be made in foundational theories and cutting-edge technologies of artificial intelligence (Tan Tieniu, 2019). Professor Geoffrey Hinton from the University of Toronto, Canada, has insisted on researching deep neural networks for 30 years, ultimately making deep learning the key to the revival of artificial intelligence. Hinton’s student, Ilya Sutskever, co-founder and chief scientist of OpenAI, continued to delve into his teacher’s research, leading to the eventual public release of ChatGPT. China’s research in artificial intelligence technology has a problem of being application-heavy and foundational-light, focusing more on the application level while neglecting foundational research and theoretical innovation. There is a need to further strengthen R&D in foundational fields and enhance original capabilities.

First, strengthen research in foundational fields.Foundational fields include frontier foundational theories, underlying technologies, key common technologies, and foundational software and hardware. The state should establish special research platforms to focus on supporting research in core devices, key equipment, high-end chips, and foundational software. Second, promote the development of disciplines related to artificial intelligence technology. Facilitate basic research in relevant disciplines such as brain science, cognitive science, mathematics, and psychology. Study the legal, ethical, and other issues arising from the application of artificial intelligence technology. Encourage interdisciplinary research to enhance innovation capabilities and promote the generation of original results. Third, promote the open sharing of data resources. China’s foundational research, technology, and industrial fields exhibit a fragmented state, unable to leverage comprehensive advantages, and should set up institutions integrating production, education, and research, forming unified goal orientation and sharing core technologies and data resources.

(2) Promote Open and Inclusive International Cooperation

Artificial intelligence is an industry requiring high levels of international cooperation. Currently, leading multinational companies in the field of artificial intelligence, such as IBM, Google, and Microsoft, are all led by managers born in India. The origin of ChatGPT can be traced back to the 2012 ImageNet competition, founded by Fei-Fei Li, who was born in China. The winning team, AlexNet, was led by Professor Hinton from the UK and included two students born in the Soviet Union, whose main work in the competition was to extend the convolutional neural network theory of French scientist Yann LeCun (Gu Xianfeng, 2016). Although China is currently facing technological restrictions and trade blockades from Western countries such as the United States, it should still adhere to the approach of “going out and bringing in” to promote international cooperation in artificial intelligence.

First, widely participate in international artificial intelligence-related activities.Actively promote joining global and regional international organizations or agreements related to artificial intelligence, and proactively align with relevant rules, regulations, management, and standards to formulate implementation plans for domestic reforms in related fields. Actively participate in international artificial intelligence-related conferences, competitions, and other activities to deeply communicate and exchange relevant research results. Explore the establishment of governance rules around major issues such as cross-border data flow, and actively participate in the rule-making of the global artificial intelligence field to contribute “Chinese wisdom” to the development of global artificial intelligence. Second, build a global service platform. Fully optimize the business environment, strengthen the protection of intellectual property and the legitimate rights and interests of foreign investment, and attract multinational companies in the field of artificial intelligence to invest in China. Establish multi-level artificial intelligence service platforms involving governments, enterprises, and research institutions to attract global innovative resource elements to participate in the development of China’s artificial intelligence technology and industry. Improve policies to promote trade in artificial intelligence industries and strengthen institutional supply and legal guarantees. Third, open and share artificial intelligence technology. Build multi-level open-source platforms for artificial intelligence, promote open innovation methods in artificial intelligence, encourage artificial intelligence technology development companies to open APIs (Application Programming Interfaces) to facilitate developers in calling foundational models to develop applications. Establish a transmission mechanism for foundational common standards and industry application standards in artificial intelligence to streamline the artificial intelligence application ecology.

(3) Strengthen a Layered and Comprehensive Talent Cultivation and Introduction

With the emergence of ChatGPT, Google, Baidu, Tencent, and Alibaba have all announced that they will soon release similar related applications, indicating that competition in the field of artificial intelligence will become even more intense in the coming years. Based on the general intelligence and deep understanding attributes of ChatGPT, it can be anticipated that some current job positions will soon disappear. According to predictions by Forbes magazine, artificial intelligence may take over nearly 1 billion jobs globally in the next decade. The key to developing artificial intelligence lies in talent, and the shortage of mid to high-end talents has become a major bottleneck for China to grow stronger in artificial intelligence. Additionally, the public’s understanding of artificial intelligence technology needs to be further improved; everyone needs to adapt to the technological wave of the artificial intelligence era.

First, strengthen the cultivation and introduction of high-end technical talents.Establish a multi-level mechanism for cultivating and introducing leading talents in artificial intelligence within the government and enterprises, focusing on cultivating innovative entrepreneurial talents in artificial intelligence for technological innovation and industrial development. Second, strengthen education related to artificial intelligence. Introduce artificial intelligence courses at the compulsory education stage and gradually promote education for all. Place a high emphasis on foundational work in the field of artificial intelligence education, strengthen the construction of high-level educational content and resource platforms, and accelerate the training of teaching staff for artificial intelligence-related courses. Third, strengthen re-education and training. Establish a re-education and training mechanism for the labor force, enhancing education and training in fields related to artificial intelligence technology to improve the re-employment capabilities of the labor force.

(Author: Wang Qinghua, Deputy Director of the Academic Office of the National Information Center; Source: “China Price” 2023, Issue 4)