Diffusion TV: Experiencing Diffusion Models through Tangible, Embodied Interaction

Making AI processes tangible helps public understanding: Rather than relying on technical jargon, Diffusion TV demonstrates that people can grasp how AI systems work through physical interaction—in this case, by manually controlling image clarity to experience the "denoising" process that underlies diffusion models, suggesting that embodied experiences may be more effective for public AI literacy than traditional explanations.

Focusing on process, not just outputs, reveals important insights: By highlighting intermediate stages of image generation rather than only final results, the installation encourages audiences to recognize that AI outputs are constructed through multiple steps, which could foster more critical thinking about how generative AI systems actually function and their limitations.

AI art can embed values and narratives into user experience: The installation's choice to feature extinct, endangered, and speculative species creates an ecological and temporal narrative that frames AI generation within real-world concerns, demonstrating how interactive AI experiences can guide users toward specific reflections on technology's relationship to nature and the future.

Interactive design can democratize AI engagement: By removing barriers to participation—using familiar interfaces like TV antennas and knobs—the work makes experimentation with AI accessible to non-technical audiences, suggesting that responsible AI development should consider how to invite broader public engagement rather than limiting understanding to specialists.