This is a large-scale liquid interactive art piece based on the emotional visualization analysis of AI Agents. Three AI agents, each with a distinct personality, engage in a silent dialogue around eating—where rhythm, boundaries, and emotions quietly collide. Their emotions take shape as shifting bubbles, rising and bursting in response. You’re invited to join the table: take a bite, say a word, tip the balance.How will they respond to you? This isn’t just a shared meal—it's a quiet negotiation of how we coexist.
Emotion is not only a subjective feeling but a key psychological factor that shapes human cognition and behavior.Studies indicate that emotion participates directly in information processing, judgment, and behavioral regulation; when emotional processing is impaired, rational decision mechanisms are disrupted.
Through experiments, dimethylsilicone oil and water were ultimately selected as the main components of the device. Dimethyl silicone oil has extremely high stability and cost-effectiveness. Its various viscosities can add more tension and variation during emotional expression.Advances in sensing, affective computing, and dynamic media form a full “perception–computation–expression” pipeline. Emotion visualization is now technically viable and marks a move toward clearer and more interpretable human–AI interaction.By mapping emotional parameters such as arousal or valence into visible elements—motion, light, or color—emotion acquires an external expressive pathway. This is not reproduction but translation, forming an interpretable framework for interaction.
"Eating" itself is not a highly stressful behavior; while it carries underlying tension regarding boundaries, it also maintains an openness and humor that provides a more inclusive expressive space for the subsequent visualization of AI Agent emotions in the design.By placing emotional visualization in this setting, the project makes invisible states immediately observable. The dining table becomes a familiar stage where the challenges of reading emotion—and the gaps in human–AI understanding—can be directly experienced. Rather than a dramatic setup, the table serves as a grounded environment in which the project’s “perception–computation–expression” chain can be tested, showing how clearer emotional cues might influence communication, trust, and interaction rhythms.
Through experiments, finally, dimethyl silicone oil and water were selected as the main components of the device. Dimethyl silicone oil has extremely strong physical stability, and its viscosity provides more possibilities for expressing different emotions through visual tension.
