Challenges, Risks, and Opportunities: Marxist Political Economy Review of AIGC ^†

Abstract

AIGC is a hot topic at present, and its display of human-like creativity has led to a rethinking of “labor” In order to unravel the mysteries brought about by new technology, this article takes the research perspective of Marxist political economy and, by sorting out the concepts, evolution, and application cases of AIGC, comes to understand what AIGC is. It analyzes the argument that AIGC poses a challenge to the composition of labor value, responds with the basic theory of Marxist political economy, and further examines the so-called labor of AI. It is worth noting that AI, as a revolutionary technological tool, has many worrying risks in practice. But it is not a terrible innovation. Just as Marx’s attitude towards machinery replacing handicrafts, new technology, and new tools always bring new opportunities for human development.

1. Introduction

In November 2022, Open AI’s ChatGPT was released and instantly became popular across the web. The chatbot can not only have high EQ and IQ conversations, but also automatically generate copywriting, novels, scripts, papers, etc., promoting human–machine interaction [1]. The breakthroughs in algorithms and computing power, as well as the support of massive data, have enabled AIGC to achieve a qualitative breakthrough since 2022. Machines have begun to use “imagination” to replace some of the mental labor in past industries, and machines can even generate similar products more efficiently than humans. At this point, does Marx’s Labor Theory of Value’s conclusion that “all value comes from human labor” still hold true? This article hopes to explore and research this issue.

2. The Evolution and Practical Application of AIGC

Among the hot topics about AIGC, there is no shortage of exaggeration through sensationalism to gain traffic. To evaluate this emerging phenomenon calmly and correctly, it is necessary to have an objective and clear understanding of its concept, evolution, and application.

There is currently no uniform standard definition for how to define this concept in either academia or industry. An intuitive understanding of AIGC is the creation of content with the help of artificial intelligence algorithms [2] (15–23). From a technical perspective, AIGC refers to a technology that uses artificial intelligence technologies such as GANs and large pre-trained models to find patterns from existing data and generate related content through appropriate generalization capabilities [3]. From a production perspective, China’s industry and academia tend to view AIGC as “after Professional Generated Content and User Generated Content, a new production method using artificial intelligence technology to automatically generate content.” It is a type of content classified from the perspective of content producers, a way of producing content, and a collection of technologies used for automated content generation [4].

Tracing back to the many technological innovations that triggered the explosive growth of content generated by artificial intelligence, representative technological changes include GAN (Generative Adversarial Networks) that emerged in 2014 [5] (139–144). In short, the purpose of generative model G is to generate large amounts of “fake content” to deceive discriminative model D until discriminative model D has difficulty determining whether generated content is real or not while generating “realistic” content through mutual play between two models. It can be seen that thanks to algorithmic breakthroughs, GAN’s mature application has laid the foundation for core technological development for content generated by artificial intelligence.

3. Responding to AIGC’s Challenge to Labor Value Formation

The challenge posed by AIGC to the labor theory of value stems from the fact that the production process of AIGC seems to mimic human creative activity, replacing human mental labor, which is undoubtedly worrying. The rapid development of AIGC has brought people back to Marx’s Labor Theory of Value.

The emergence of “unmanned” production makes it difficult to measure the value of AI-created products with human labor. After all, if the production of goods can be separated from human labor, why should the value of goods still be determined by human labor? Marx’s Labor Theory of Value seems to have failed in front of AIGC. Scholars Hu Bing and He Yunfeng from Shanghai Normal University believe that Marx mentioned in Capital that human labor has acquired a new form of labor in machine production, namely “the new labor of using eyes to watch machines and using hands to correct machine errors” [6] (431), that is, machine production cannot be separated from human participation. In their proposed weak artificial intelligence stage, technological progress will further drive human labor out of machine institutions. The characteristic of weak artificial intelligence technology is that it can autonomously collect information, make simple judgments, and interact with other artificial intelligence or humans based on this information. This characteristic can solve the new problem caused by machine production—“watching machines and correcting their operating errors” so machine institutions can completely get rid of the shackles of human labor with the help of weak artificial intelligence technology and achieve higher production efficiency. Artificial intelligence represents high-efficiency production, which makes “replacing human labor” a trend. This trend must have a goal, which is to establish unmanned factories. If mainstream social material production exists in form of unmanned factories, then it is diametrically opposed to insisting on explaining production problems based on human labor [7] (5–14). Their views are concentrated on living labor as no longer the only source of value creation, and the subject of labor is not necessarily human beings, so the classical Labor Theory of Value is no longer suitable for the era of artificial intelligence.

In the eyes of staunch defenders of the Labor Theory of Value, AIGC’s emergence has not shaken Marx’s basic understanding of commodity value formation; AI is still a tool for productive human labor.

Firstly, human labor has not disappeared in the production process. When AI-generated content rivals human creation, a habitual understanding is that AI has acquired a human way of thinking. The reason adversarial models can judge “properties” for content is inseparable from a new profession—AI annotators. The development of AI depends on algorithms, computing power and data; human annotation labor generates data used for training; it is humans who train AI to understand new things; then AI generates similar products based on this understanding using its “imagination”

Secondly, human labor is hidden in new divisions of labor. Marx once discussed the relationship between labor power and machinery in Capital. For workshop handicrafts, “cooperation based on division of labor achieved its typical form in workshop handicrafts” during the machinery industry period, “the skill in using tools as well as tools themselves were transferred from workers to machines; tool efficiency was liberated from restrictions on human physical strength; thus, technical basis for division of workshop handicrafts disappeared” [6] (390–483). Under this change, a new division of labor emerged; just as people left manual tools to operate machinery, humans gradually left previous positions to participate in production activities in another way with which we are not yet familiar.

Finally, value is still undifferentiated by human labor congealed in commodities. However, the value of AIGC commodities is not higher than the value of commodities produced by traditional technology solutions; the value of similar commodities is still determined by abstract labor.

4. Beware of Multiple Risks Brought by AIGC

AI technology has not made machines self-conscious workers; it is still a tool for productive human labor. However, it is worth noting that as a revolutionary technological tool, many worrying risks have been exposed during its practice.

The emergence of AIGC has worsened the working situation for some employees. For the profession of AI annotator, the rapid development of AIGC has had another impact on them. The job of an AI annotator is to train AI’s “brain” using traditional, “non-AI” methods, and their simple, repetitive, and heavy data annotation work is very similar to the scene in “Modern Times”, where Charlie Chaplin tightens screws. At the same time, since this kind of labor does not require special professional skills, it is often low-paying hard work. What is even more regrettable is that for the current technical path of AIGC, this kind of manual annotation work is unavoidable, and high-intensity alienated labor has become an indispensable part of AIGC application before technology makes another breakthrough. Subsubsection

Revolutionary AI technology has also brought a series of intense impacts on the labor market. Firstly, it has caused serious “skill devaluation” for technical professionals engaged in related creative work. Secondly, AIGC’s high efficiency also reduces the demand for labor in related industries. Finally, as a result of a series of events, an oversupply of labor goods in certain industries is bound to occur in a short period of time, and labor prices (wages) have stagnated or even decreased.

AIGC may exacerbate data-based monopolies. Under similar algorithms and computing power conditions, the amount of data resources determines the quality of AIGC; these data include process data generated by users during generation feedback and adjustment when using AI applications. Users will only choose the smartest AI in the current application scenario; no one cares who is second; under the “Matthew effect”, user number advantage becomes a source of quality advantage for AI applications. Therefore, companies with large numbers of users based on occupying user data can achieve business advantages over other similar companies. So, when users expect AI tools to be more powerful, they also unconsciously expect the emergence of data oligarchs.

5. Seizing New Development Opportunities Brought by AIGC

AIGC is not a terrible innovation. Like other revolutionary technologies, their potential application risks can be avoided through proactive actions. At the same time, accepting this new technology can bring new opportunities for human development.

For creators using AI tools requires an adaptation process. The high efficiency of AIGC is reflected in the ability to generate a large number of similar products within a short time based on prompt language. A way of thinking is to use AI tools to generate drafts, sketches, or summaries to provide inspiration and assist creation which effectively avoids the problem of AI’s lack of accuracy; at this time, AI tool is not a substitute but an assistant.

AIGC not only assists thinking but also frees people from some tedious and necessary work; fully utilizing AI tools will give people greater motivation for creative eruption. In the medical field, AI also shines. Biologists from the University of Washington School of Medicine, through training AI, can generate a large number of brand new protein structures based on certain rules even if these proteins do not exist in nature, then, through continuous iteration and optimization, obtain stable proteins that can achieve specific functional goals providing design drawings for later manufacture and research in the laboratory [8].

As a powerful tool, AIGC has created new opportunities for promoting the comprehensive development of people. Firstly, the application of AI has raised the upper limit of human ability. The increasing maturity of AIGC has provided people with new technical paths to solve problems, such as designing new proteins, which has become possible under the latest AI models. High efficiency brings low cost, allowing people to use limited resources more economically to achieve production labor. The power of using AI is precisely the powerful ability that humans show when using it to solve problems. Secondly, the application of AI has improved the basic level of human ability. Based on this cross-modal technical application, real-time translation between different languages has become a reality. Existing AI technology can also solve corresponding problems for disabled people through methods such as picture-to-speech, speech synthesis, and optical character recognition. And research is already underway to achieve the transformation of brainwave signals into neural electrical signals through AI technology, and then help paralyzed patients walk again. Finally, the application of AI has won more time for human innovation labor. Looking back at history, the improvement of the industrial mechanization level has given more people the opportunity to be freed from physical labor and move towards positions mainly based on mental labor, which further promoted the improvement of productivity levels. The popularization and application of AIGC are also conducive to freeing people from tedious, boring, and non-innovative mental labor, allowing people to have more time to focus on innovative labor that tools cannot replace.

6. Conclusions

Marx once commented on new technologies and new tools in Capital: “Since steam and new tools have turned old workshop handicrafts into large-scale industry, under the leadership of the bourgeoisie, production power has developed at an unprecedented speed and scale” [9] (655). Although there were many hostile voices towards new technologies and new tools among workers’ complaints in the 19th century, in Marx’s view, new technologies, and new tools were not culprits; it was still only capitalists who pursued profits that caused workers’ miserable experiences. On the contrary, it was new technologies and new tools that promoted the development of productive forces and social change. Obviously, AIGC is not a “Pandora’s Box”, but like many revolutionary technologies in the past, it is a “double-edged sword” that we need to study carefully. It is sharp but controllable.