2026
Video, generative AI, computer screen, Industrial Angle Steel Shelving
dimension variable, 09m 20s
In this work, rather than physical site-visitation, I adopted a remote approach—harvesting secondary and tertiary historical memories from the web. By collecting archival data of Taiwan’s Cold War military ruins and urban exploration records from online forums—including data from the Xizihwan Chiang Kai-shek Underground Command Post and abandoned missile bases on Mount Chai—I fed these fragments into a generative AI model. However, during the computational process, the contemporary biases inherent in the AI model surfaced: it failed to distinguish between the weathered concrete of actual war vestiges and the high-resolution textures of First-Person Shooter (FPS) games. Within the logic of the algorithm, these dark, humid Taiwanese underground spaces—shrouded in moss and tangled pipes—were systematically misread as level designs from games like Counter-Strike, Call of Duty, or Half-Life. Consequently, the generated imagery exhibits eerie distortions, with digital noise resembling weapons, cameras, or severed limbs emerging at the bottom of the frame. History is forcibly filtered through the visual syntax of a shooter game. The installation mimics the aesthetic of a "gold farm" or a cryptocurrency mining rig, utilizing a multi-screen and server-rack setup. In the video, the AI-reconstructed Taiwanese military bases become a self-generating, infinite copy—endlessly reassembling and expanding in a loop that never finds an exit.







