These symbols are output from a StyleGan-2 neural network trained on the contents of Carl Liungman's Dictionary of Symbols.
Inputs range from pictographs to geometric designs and simple characters. The images themselves are cultural signifiers sprawled across the boundaries of language, humming with layers of broad and diverse information. The neural net produces output at once full of readable meaning and also hollowed out of any cultural context.
These forms, pinned up like specimens, show the AI synthesizing visual information with no thought given to utility, communication, or charm. But when seen together they communicate, inevitably, their origins as objects of human intention.