Abstract:

Since the technological revolution triggered by ChatGPT in 2022, generative artificial intelligence large models have pushed digital production into a new stage of digital and physical transformation. With its intensity comparable to the machine revolution of the industrial age, this transformation has given rise to AI-capitalism. Taking the dual capital composition of core architecture as fixed capital and dynamic data as circulating capital as their carrier, large models are essentially technological entities of digital materialized labor. This existence is neither an independent value-creating entity nor a technological autonomous entity; rather, it is a tool through which capital realizes its reproduction by reconfiguring the process of digital labor. Digital labor is the main reliance for human-machine interaction, and productive digital labor is the direct source of value creation. The production of large models enhances the capital organic composition. Embedded in the capitalist value reproduction system, this production further expands the source of surplus value. The high value-added of large models is deceptive; the value that seems to be autonomously produced is actually the old value of the core architecture and dynamic data of digital intelligence, as well as the value created by the physical labor of digital intelligence workers through the addition of new value. The source of surplus value has always been the living labor of digital workers, and the core law that “labor is the sole source of value” has not fundamentally changed due to the technological form of large models.

Key words: AI-capitalism, large models, digital labor, Marx’s labor theory of value, organic composition of capital