Blue Noise
Two Channel audio, video, Spectrograms
(Part of Museum of Blue)
The work comprises of two audio spectrograms. One renders a neural network–generated interpretations of "blue," produced using the XLNet language model, which was trained on a corpus of over 32.89 billion words. As a result, these interpretations synthesize a vast multitude of textual experiences of blue into a single, composite voice. The other spectrogram draws on responses from the elderly women from my village Dhokri in Pakistan, an "analogue" public whose interpretations remain inaccessible to AI. Many women just laughed at the absurdity of my question which was, ‘what does the color blue mean to you’. These responses are emotional and contextual, shaped by memory, association, and lived experience. I am focusing on the voice of collective intelligence in relation to the synthesized voice of artificial intelligence. These voices carry information, as human voices are not static - they shift and change, and reveal context, while the machine generated voice is full of audible knowledge, yet feels static.