Unmake Lab and Foreseen Agency discussed how their respective artistic practices are influenced by socio-economic structures in South Korea and Hong Kong. The conversation focused on the similarities and differences in their methodologies, how their work decentralises knowledge production, and how these processes continually feed back into their artistic and curatorial endeavours.

Reevaluating Ecological Extraction through Datasets

The final session of the day featured a dialogue on artistic methodologies, beginning with a presentation by Binna Choi from the Seoul-based collective Unmake Lab. Titled “Data Setting Beyond the Future,” Choi’s discussion explored how artificial intelligence and datasets can function as lenses to observe society rather than merely tools to generate content. She shared the story of a broken stone retrieved from a controversial, human-made sand mountain in South Korea—a landscape dramatically altered by river mining and construction. By creating a speculative dataset from these damaged stones, Unmake Lab trained an object detection AI that continuously misidentified the broken stones as “fresh” commodities. This intentional framing challenged the predictive nature of AI, turning it into a reflective medium that highlights humanity’s insatiable drive for ecological extraction and the cascading consequences of human intervention.

AI as an Introspective Tool for Climate Action

Expanding on the theme of climate-related disasters, Choi detailed another layer of their research involving endangered mountain goats whose habitats had been devastated by wildfires. Unmake Lab utilized archival black-and-white trail camera footage collected by wildlife conservationists to train a new AI model. During the machine learning process, they discovered that using a pre-trained commercial AI model resulted in generated creatures featuring “friendly eyes”—a byproduct of the AI’s prior training on domestic companion animals. The AI also struggled to comprehend the random, atypical poses captured by the unmanned cameras, preferring frontal or side views standardized by human vision. Through this process, Choi illustrated how datasets inherently project a human-centric worldview, while the ghostly, failing generations of the AI served as a more genuine portrait of the fragile environments these endangered species inhabit.

Defragmenting Urban Memory and Infrastructures

Following Unmake Lab, Shan Wong of Foreseen Agency shared insights into their speculative design methodology, framing their artistic practice through the computational concept of “defragmentation”. Wong detailed “Ding Lab,” a deeply participatory project where the artist duo embedded themselves in a Hong Kong tram depot for months of field research. Engaging directly with the workers, they uncovered the forgotten history behind the iconic mandatory green paint originally used during the British colonial era. To explore this erasure of collective memory, Foreseen Agency created an installation acting as a tram cyborg’s visual cortex, which replaced real-world green with transparency voids and repainted them using generative archival memory. The artwork critically examined how historical narratives in Hong Kong are continuously revised and smoothed over to fit new agendas, asking what remains of underlying identities when physical evidence is painted over.

Speculating the Future of Vacant Spaces

Concluding the presentation, Kachi Chan of Foreseen Agency introduced their most recent installation responding to the contemporary socio-economic landscape of Hong Kong. Reacting to the surge of empty storefronts in traditionally busy districts post-Covid, the duo arranged sixteen flatbed trolleys—objects typically associated with grassroots labor—on the exhibition floor. Chan explained how they scraped internet leasing data to build virtual 3D spaces based on the floor plans of approximately 1,600 vacant stores. Utilizing AI noise as a foundational aesthetic and technical element, the artwork visualized the digital spaces from the speculative, non-human perspective of the “jobless” trolleys aimlessly navigating these empty shops. By merging raw data with spatial generation, Foreseen Agency highlighted the invisible logics of capitalist structures and the stark realities facing marginalized labor in the city.