AI chips power the development and deployment of AI at scale. Photo: TUCHONG

In 2018, E Weinan, Chinese mathematician and academician of the Chinese Academy of Sciences (CAS), proposed for the first time the concept of “AI for Science” to the international academic community, defining it as scientific intelligence. According to Tang Chao, Chinese biophysicist and CAS academician, the “AI for Science” initiative aims to examine the application of machine learning in various fields of science and engineering, with “science” referring to the natural sciences. Extending this framework, the application of AI in humanities and social sciences can be termed “AI for Humanities and Social Science” (AI for HHS).

Artificial intelligence, an advanced form of machine intelligence created by humans, finds a unique application in the humanities and social sciences (AI for HSS). This interaction between human and machine intelligence transforms the paradigms of knowledge innovation and accumulation, explores the diverse facets of the human spirit, and tests or reveals fundamental social mechanisms. Scholars have learned from the deployment of AI for Science that the development of AI for HSS necessitates at least three key pillars.

The first pillar is AI-aided data integration. Data is the fuel for machine intelligence, which can be activated with the data trails of human thought and actions in addition to knowledge accumulated by humans. Since the advent of digital sensors and terminals, human thought and actions have been accompanied by data production and aggregation, providing an excellent opportunity to improve the efficiency of AI and fully leverage its capabilities. Transforming vast amounts of historical and contemporary multimodal material, such as text, images, speech, music, dance, rituals, survey results, and statistics into fine-grained, computable data is beyond human capabilities, making collaboration between human and machine intelligence the sole viable option.

The second is AI-based factual analysis. Human knowledge innovation and accumulation have undergone three paradigm revolutions, which the author refers to as the academic paradigm, the disciplinary paradigm, and the intelligent paradigm. The disciplinary paradigm, currently represented by discipline-specific education and research, emerged from the scientific revolution beginning in the 16th century and revolves around the use of quantitative methods to study the laws of nature and human society. When human intelligence is incapable of dealing with massive amounts of data, algorithms can partly replace quantitative methods. The merit of algorithms empowered by machine intelligence lies in their capacity for incorporating connections between plenty of objects to test hypotheses and discover patterns amid a multitude of complex correlations.

The third pillar is AI-driven theory building. Considering the infinite nature of human thought and actions, the exploration of the human spirit and social patterns is also infinite. However, within a given time and space, these elements are relatively stable and can be considered finite, and thus can provide temporarily stable scenarios for theory building. Due to the limitations of human intelligence, interconnected objects are artificially categorized as subject matter of different disciplines within the disciplinary paradigm, expressed as fragmented, dispersed knowledge with complex presuppositions. Interaction with machine intelligence enables human intelligence to penetrate deeper into the human spirit and social patterns amid complex and organic correlations, which requires a much higher level of imagination compared to the disciplinary paradigm.

The goal of developing AI for HSS is to empower human intelligence with the help of machine intelligence, rather than merely finding application scenarios for AI. On one hand, this allows the innovation and accumulation of humanities and social science knowledge to align with the organic nature of the human spirit and social phenomena. On the other hand, the popularization of this knowledge serves to meet people’s demand for knowledge in production and everyday life, thereby facilitating societal development.

Qiu Zeqi is a professor from the Department of Sociology at Peking University.

Edited by WANG YOURAN