王可欣 & 方铭璐

《光协议：编译生命》是一件由三部同步影像构成的生物艺术作品，呈现生命如何被AI预测、被用于校准传统算法，并最终成为一种以身体完成的计算。

左侧影像将真实毛霉菌生长与一个在真实真菌数据上训练的神经元胞自动机模型并置，使AI预测与生命实际生长之间的偏差变得可见。中央影像记录多头绒泡菌对激光的真实反应，其速度、方向与网络变化被转化为数据，用于校准基于规则的传统算法。右侧影像则将食物设为节点、光设为约束，让黏菌与算法面对同一道问题：算法生成抽象路径，黏菌以真实生长的身体作答。

三部影像由此构成一个从生命到模型、再从模型返回生命的循环：AI预测生命，生命校准算法，最终两者进行对比求解。作品反转了“AI训练生命”的想象：机器同样需要被生命训练，并追问当生长被压缩成数据、参数与答案时，我们是在理解生命，还是仅仅建立了一个便于计算的生命模型？在AI的预测、算法的规则与黏菌的身体之间，究竟哪一种才算智能？

Light Protocol: Compiling Life is a bio-art work composed of three synchronized video channels, tracing how life is predicted by AI, used to calibrate a conventional algorithm, and finally becomes a form of computation carried out by the body itself.

The left channel places the growth of living Mucor mold alongside a Neural Cellular Automata model trained on real fungal data, making visible the divergence between what the AI predicts and how life actually grows. The central channel records the real responses of Physarum polycephalum to laser light; its speed, direction, and shifting networks are translated into data used to calibrate a rule-based conventional algorithm. The right channel sets food as nodes and light as constraints, posing one and the same problem to both the slime mold and the algorithm: the algorithm generates an abstract path, while the slime mold answers with a body that has to actually grow.

The three channels thus form a loop that runs from life to model and back from model to life: AI predicts life, life calibrates the algorithm, and in the end the two are set against each other to solve the same problem. The work inverts the familiar image of “AI training life”: the machine, too, must be trained by the living. It asks whether, once growth has been compressed into data, parameters, and answers, we are understanding life itself, or merely building a model of life that is convenient to compute. Between the predictions of AI, the rules of the algorithm, and the body of the slime mold, which one counts as intelligence?

致谢：感谢曳尾菌（周晴烽）提供本作品所使用的黏菌（多头绒泡菌）菌种。

Acknowledgment: We thank Zhou Qingfeng (Yeweijun) for providing the slime mold (Physarum polycephalum) cultures used in this work.

王可欣Kexin Wang，跨学科艺术家、研究者与教育者。毕业于罗德岛设计学院与布朗大学(双学位硕士)，现为 MIT CSAIL 附属研究员、MIT Media Lab 科技艺术黑客松核心组织者。实践横跨数字生命、人工智能与数字制造，关注复杂系统演化的模拟与预测。作品曾于威尼斯、奥克兰、东京、中国香港、波士顿等地展出，并发表于 SIGGRAPH Asia、CHI、UIST 等国际学术会议。现任 Artineer Lab 高级研究员，FunguyLab创始人。

方铭璐Minglu Fang，跨学科设计师、研究者与策展人。本硕毕业于中国美术学院现就读加州大学伯克利分校。实践横跨空间、叙事、数字媒体、交互装置与策展，关注人工智能、生物混合系统及人与非人主体的共生关系。曾参展威尼斯建筑双年展、上海双年展、深圳设计周等。现任 Artineer Lab 研究助理，FunguyLab艺术家。

Kexin Wang is a multidisciplinary artist, researcher, and educator. She graduated from the Rhode Island School of Design and Brown University (with a double-degree master's degree), and is currently an affiliated researcher at MIT CSAIL and the core organizer of the Technology Art Hackathon at MIT Media Lab. Her practice spans digital life, artificial intelligence, and digital manufacturing, focusing on the simulation and prediction of complex system evolution. Her works have been exhibited in Venice, Auckland, Tokyo, Hong Kong, and Boston, and have been published at international academic conferences such as SIGGRAPH Asia, CHI, and UIST. She is currently a senior researcher at Artineer Lab and the founder of FunguyLab.

Fang Minglu is a multidisciplinary designer, researcher, and curator. She graduated from the China Academy of Art with a bachelor's and master's degree. Currently, she is studying at the University of California, Berkeley. Her practice spans space, narrative, digital media, interactive installations, and curating, focusing on artificial intelligence, bio-hybrid systems, and the symbiotic relationship between humans and non-human entities. She has participated in exhibitions at the Venice Architecture Biennale, Shanghai Biennale, and Shenzhen Design Week. She is currently a research assistant at Artineer Lab and an artist at FunguyLab.

光协议：编译生命

网站 / 社交媒体

关注议题 · AI Topics