Social Sciences in China (Chinese Edition)

No. 5, 2023

The Nature of AI and Its Limits in the Light of the Opacity of Machine Cognition

(Abstract)

Dong Chunyu

One of the core theoretical issues in current AI research is understandability. Most computational simulations do not allow one to access transparency in the traditional epistemological sense. The enormous number of complex technological calculations in the implementation of computing science make it impossible to review the entire process of calculation, leaving a blind spot in human understanding. Based on the distinction between the opacity of cognition and the opacity of epistemology, the opacity of machine cognition will lead to a lack of understanding. An analysis of the possible degree and requirements of human ability to understand machines in the light of epistemological transparency allows us to conduct a relatively accurate in-depth discussion on the nature and limits of machine understandability and clarify the characteristics of machine epistemology. Thus this article seeks to provide a heuristic pathway to the understanding of the essential connections and differences between artificial intelligence and human intelligence.