Who Chooses the Artwork? Curatorial Agency and Distributed Intent in Botto
论文探讨了Botto作为分布式策展系统，通过生成、排名、投票等环节形成循环，分析了其内部模型与社区参与者如何共同影响艺术作品的选择和认可。
0 citationsRead paper
Selected work
Representative Papers
Disclosure and dissolution: explainability, AI power, and situated agency in understanding
论文通过赛博格理论、宇宙技术学和故障女权主义探讨AI的政治与哲学意义，提倡可解释AI及社区参与来促进多样性和伦理创新。
0 citationsRead paper
Govern the Model, Not Only the Data: Storage, Circulation, and Learning in Creative AI
论文探讨了联邦学习中的模型治理问题，提出创意社区应在存储、流通和学习层面进行治理，并提出了四个设计原则以实现对模型及其联合的治理。
0 citationsRead paper
Social Facilitation of Creative Reflection: AI-agents and Humans
This study investigates whether AI agents can replicate the social facilitation effects observed in human creative reflection, with a particular focus on how nonverbal interactions influence creative output. Grounded in social facilitation theory and integrating human–computer interaction design with AI agent simulation techniques, the research establishes a controllable collaborative experimental environment to systematically examine how an AI partner’s nonverbal behaviors modulate human performance in creative reflection tasks. The project introduces the first theoretical framework of social facilitation tailored to human–AI co-creativity, empirically demonstrating the critical role of nonverbal interaction in human–AI collaboration. These findings provide both an empirical foundation and conceptual guidance for the future development of intelligent creative systems, informing both theoretical modeling and practical design.
0 citationsRead paper
Vertigo Vertigo: Reconstructing a Cinematic Ideal through its Predictive AI Double
This study investigates the extent to which generative AI internalizes classical cinematic conventions by reconstructing Alfred Hitchcock’s *Vertigo* scene-by-scene using only 2.78% of its keyframes as input to a large video diffusion model, which interpolates between start and end frames. Integrating keyframe anchoring, computational analysis, and media-theoretical critique, the work extends the film’s theme of “artificial ideal reconstruction” to the ontological level of the medium itself, employing AI generation as a probe into the model’s mechanisms of compressing and reproducing cinematic language. Experimental results demonstrate that 73.1% of reconstructed frames were judged as plausible reconstructions, with only 3.6% exhibiting severe distortion, confirming that canonical filmic norms are deeply embedded within the model’s priors.
0 citationsRead paper
Recent publications
Latest Papers
Who Chooses the Artwork? Curatorial Agency and Distributed Intent in Botto
论文探讨了Botto作为分布式策展系统，通过生成、排名、投票等环节形成循环，分析了其内部模型与社区参与者如何共同影响艺术作品的选择和认可。
0 citationsRead paper
Disclosure and dissolution: explainability, AI power, and situated agency in understanding
论文通过赛博格理论、宇宙技术学和故障女权主义探讨AI的政治与哲学意义，提倡可解释AI及社区参与来促进多样性和伦理创新。
0 citationsRead paper
Govern the Model, Not Only the Data: Storage, Circulation, and Learning in Creative AI
论文探讨了联邦学习中的模型治理问题，提出创意社区应在存储、流通和学习层面进行治理，并提出了四个设计原则以实现对模型及其联合的治理。
0 citationsRead paper
Social Facilitation of Creative Reflection: AI-agents and Humans
This study investigates whether AI agents can replicate the social facilitation effects observed in human creative reflection, with a particular focus on how nonverbal interactions influence creative output. Grounded in social facilitation theory and integrating human–computer interaction design with AI agent simulation techniques, the research establishes a controllable collaborative experimental environment to systematically examine how an AI partner’s nonverbal behaviors modulate human performance in creative reflection tasks. The project introduces the first theoretical framework of social facilitation tailored to human–AI co-creativity, empirically demonstrating the critical role of nonverbal interaction in human–AI collaboration. These findings provide both an empirical foundation and conceptual guidance for the future development of intelligent creative systems, informing both theoretical modeling and practical design.
0 citationsRead paper
Vertigo Vertigo: Reconstructing a Cinematic Ideal through its Predictive AI Double
This study investigates the extent to which generative AI internalizes classical cinematic conventions by reconstructing Alfred Hitchcock’s *Vertigo* scene-by-scene using only 2.78% of its keyframes as input to a large video diffusion model, which interpolates between start and end frames. Integrating keyframe anchoring, computational analysis, and media-theoretical critique, the work extends the film’s theme of “artificial ideal reconstruction” to the ontological level of the medium itself, employing AI generation as a probe into the model’s mechanisms of compressing and reproducing cinematic language. Experimental results demonstrate that 73.1% of reconstructed frames were judged as plausible reconstructions, with only 3.6% exhibiting severe distortion, confirming that canonical filmic norms are deeply embedded within the model’s priors.
0 citationsRead paper