Playing Heaven
Remapping Early Modern Neo-Confucian Worlds with AI
In early modern East Asia, Neo-Confucianism and related intellectual traditions took shape over centuries, as scholars interpreted classical texts, debated moral and political questions, and pursued literary and aesthetic practices. Their histories extend far beyond familiar canonical figures and the taxonomies established in retrospective accounts. Playing Heaven investigates how the writings, scholarly communities, and cultural activities of thousands of lesser-known figures contributed to larger intellectual, social, and cultural movements.
Our team’s long-term goal is to examine the diversity of intellectual traditions within China and their reception and reinterpretation in Chosŏn Korea, Tokugawa Japan, the Ryūkyū Kingdom, and Vietnam, as well as wider exchanges involving Jesuit missionaries and European scholarship. The current case studies begin with intellectual connections between China and Korea and with Manchu materials from the Qing empire. To pursue these complex historical questions, we are also developing computational methods that combine transformer-based models, agentic AI, deterministic algorithms, and carefully curated knowledge bases.
A Schmidt Sciences Humanities and AI Virtual Institute ProjectA collaboration of the University of Hong Kong, the Hong Kong University of Science and Technology, and the National University of Singapore, 2026–2029
Pivot 機
The name Pivot invokes the concept of ji/ki 機, which in traditional East Asian cosmology refers to the generative workings of Heaven, the ultimate source of creativity governing the birth, growth, and perishing of the myriad beings. The term carries several layers of meaning and connotation, including mechanism, incipience, and the metaphor of a brocade loom.
Inspired by this conception of creative agency, Pivot uses specialized AI agents to assist historians in analysing historical writings, constructing successive interpretations of passages and essays, and comparing intellectual positions. Agentic AI approximates aspects of this generative capacity, enabling researchers to develop and pursue interpretations spanning large bodies of historical evidence. Each interpretation remains linked to its original sources to help scholars inspect, question, and revise the results and establish confidence in the plausibility and validity of their findings.
Traces 跡
The name Traces refers to the fragmentary evidence from which historians reconstruct intellectual and cultural formations and reconsider inherited classifications of schools of thought, intellectual lineages, and cultural movements that reflect the interpretive judgements of earlier taxonomists.
Traces uses an agentic AI pipeline to help historians reassess these formations against the writings and documented relationships of thousands of scholars, bringing together fragmentary evidence from different sources and scholarly traditions within a shared research environment.