Reflection

Reflection encapsulates an innovative approach to machine creativity and generative art that the artist Pindar Van Arman calls 'Reflective AI.'

Van Arman's work is born from the effort to teach machines to paint like a child. Since he began working with painting robots in 2006, his focus has been exploring the question: How can generative processes transcend algorithmic determinism and achieve pure emergence? The Reflective AI system leverages feedback loops and multimodal mark-making, reflecting on the canvas at each step. Over the course of thousands of marks, Reflection thinks by doing in an attempt to present a synthetic embodiment of a creative thought process.

Reflection's feedback loops happen each time the robot makes a mark – it photographs the canvas and captures the unpredictable movements of the paint such as splashes, drips and blending. These serendipitous inputs from the physical world open a window where true emergence can enter. The robot then reflects on the image by engaging with it in realtime, responding with fluid, gestural 'right-brained' marks and sharp, precise 'left-brained' marks. Over the course of thousands of marks, Reflection thinks by doing, embodying the primitive elements of an artist's creative process.

On Ethereum, Reflection is built with a unique long-form generative approach that utilizes the physical canvas. The system captures stages of the painting process, translates them to digital format, and composes them on mint. All of the artworks are fully on-chain thanks to an innovative contract that makes it possible to store them immutably on the Ethereum blockchain with extremely robust image quality.

Reflection pushes the boundaries of creative computing, showing that generative art can embody true emergence, on-chain art can be highly detailed, and AI art can touch the source of creation.