flower.pcd (WIP)
is a work-in-progress project that revisualises artificial intelligence training data. Using the Oxford Flowers 102 dataset, it converts images of flowers into 3D point cloud models. The work explores the liminal space between art and data, where glitches and digital artefacts mirror the hallucinations present in AI systems. Point clouds, typically used in industrial or forensic contexts, are transformed into ghostly and abstract constellations. The compositions are inspired by supernovae and cosmic dust clouds, reflecting AI’s processes of destruction and creation. By repurposing a dataset made for machine learning into an artistic context, the project reveals a hidden material and computational process that shapes artificial intelligence.