Gorgons is a series of sculptures created through the training of two distinct StyleGAN datasets: one composed of African masks and the other of Greco-Roman sculptures. The project seeks to hybridize these two different cultural and epistemological matrices, African and Western, as a strategy to expose the biases embedded in the datasets themselves and in algorithmic learning processes.
Among thousands of generated images, four figures were selected, two derived from masks and two from sculptures, which were subsequently reconstructed as three-dimensional models and 3D-printed. The pieces were manually finished by an art restorer, acquiring the appearance of archaeological objects from a future that never occurred.
By stressing the classificatory logic of algorithms and their regimes of representation, Gorgons reveals how artificial intelligence systems operate through statistical normalizations that produce exclusions, homogenizations, and cultural hierarchies. The project is situated within a lineage of contemporary artistic practices that interrogate the regimes of truth reconfigured by the emergence of artificial intelligence, particularly regarding the algorithmic production of images, identities, and criteria of legitimacy.