The device operates using a custom artificial neural network designed to find commonalities between sound and visual imagery. This approach is akin to understanding how the word 'dog' can simultaneously refer to both the animal's visual depiction and its bark. The neural network was trained using bespoke videos recorded in Mexico City, where each video frame was associated with its preceding second of sound, enabling the AI to predict the visual content that corresponds to the audio it detects.

The images generated by Blind Camera are abstract landscapes that incorporate recognizable elements of urban life, such as trees from parks, asphalt, public transportation, and concrete buildings, reflecting the urban soundscape of Mexico City. The device interprets every captured soundscape as belonging to a city, influenced by its urban-centric training data.

Blind Camera diverges from most AI systems that aim to minimize bias by embracing its local, urban, and specifically Mexican background. This choice allows the device to interpret sounds from a distinctly local perspective, prompting users to consider how AI might represent other non-global and underrepresented viewpoints. The project highlights the aesthetic, narrative, and conceptual possibilities offered by AI models when developed creatively by artists, foregrounding a unique intersection of technology, locality, and artistic expression.