Farming Instead of Mining (in collaboration with Aven-Le Zhou) 以农代矿（与周乐合作）
2024-2025, mixed media: printed card game set (paper, lamination, resin, boxed), fine-tuned large language model (LLM), digital interface (dimensions variable), high-resolution digital print on paper. 2024–2025，综合媒介：印刷卡牌游戏套装（纸、覆膜、树脂、盒装）、微调大语言模型（LLM）、数字界面（尺寸可变）、高分辨率数码打印。Installation views, Chronus Art Center, Shanghai. 装置照片，新时线媒体(CAC)艺术中心，上海。
Farming Instead of Mining_Strategy Matrix_Card Game. 以农代矿_策略矩阵_卡牌游戏。
Farming Instead of Mining_Strategy Matrix_Card Game. 以农代矿_策略矩阵_卡牌游戏。
Farming Instead of Mining_Strategy Matrix_Fine-tuned Large Language Model. 以农代矿_策略矩阵_微调大语言模型。
Farming Instead of Mining_Strategy Matrix_Infographics. 以农代矿_策略矩阵_微调大语言模型。
At the workshop. 工作坊现场。
At the workshop. 工作坊现场。
Farming Instead of Mining focuses on the issues of data production and infrastructure in the generative AI era. It critiques how mining metaphors like “data mining,” “prospecting,” “drilling,” and “cleaning” shape our perception of data, models, and value. By naturalizing data as a resource for one-way extraction, these metaphors rationalize high-energy, centralized, and unsustainable technological paths. Meanwhile, the pursuit of massive, all-encompassing models by major tech companies resembles the logic of industrial agricultural monocultures, which often neglect diversity, resilience, and ecological synergy. Instead, this project draws on principles from agroecological principles and situated knowledge, to propose alternative frameworks based on "cultivating" rather than “mining.” The project aims to reframe AI-related dominant metaphors and explore digital production methods that respect resource constraints and environmental concerns. Through co-creative approaches such as board games and fine-tuned models, Farming Instead of Mining addresses how AI strains global energy systems, intensifies climate pressures, and reshapes social conditions—offering ideas and prototypes that combine poetic expression with technical direction.
《以农代矿》聚焦生成式AI时代的数据生产与基础设施问题，反思“数据挖掘/勘探/钻探/清洗”等矿业隐喻如何塑造我们对数据、模型与价值的理解：它将数据自然化为可被单向提取的“矿产”，并在叙事层面为高能耗、强集中、不可持续的技术路径提供合理化依据。与此同时，大型科技公司对“大而全”通用模型的追逐，也类似于工业化农业的单一作物增产逻辑，往往忽视多样性、韧性与生态协同等系统性条件。该项目转而借鉴生态农业中的情境知识，提出以“培育”而非“开采”为核心的替代性框架：在重写AI的隐喻范式，在实践层面探索面向资源约束与环境关切的数字生产方法。项目以卡牌游戏、微调模型等为共创媒介，探讨了人工智能如何给全球能源系统带来重负、加剧气候压力并重塑社会状况，提供兼具诗意表达与技术指向的思路与原型。