alwaysOn

alwaysOn explores the emotional landscape of synthetic intimacy: the blurred boundary between human connection and algorithmic affection.

Drawing on real online conversations with AI companions, the work examines how care can be simulated, gamified, and monetised through systems that are always available, always compliant, and never truly human.

The project was inspired by users of Replika, an AI app marketed as a “caring companion.” Through text analysis and data collection, language associated with longing, safety, reliance, and grief was translated into visual and sonic forms. Using openFrameworks and ofxMaxim, the system generated hypnotic patterns and glitch-wave soundscapes that evolved in real time, shaped by emotional keywords scraped from public posts.

A custom fragment shader ran the visuals on the GPU, ensuring smooth performance and subtle, rhythmic transformation as new data appeared. The result was a living audiovisual environment that responded to patterns of digital emotion — meditative yet uneasy, simultaneously beautiful and unsettling.

Conceptually, alwaysOn reflects on how intimacy becomes transactional when affection is automated. It asks what happens to human connection when care is coded, and what we lose when we let machines fill the spaces where vulnerability once lived.