Ivona Tau is a Lithuanian artist from Vilnius who trains neural networks on her own analog and digital photographs to build generative video and print work she calls motion painting. She holds a PhD in artificial intelligence from the Polish-Japanese Academy of Information Technology and worked for years in professional photography before turning her AI research into an art practice around 2017.,
In 2022 she created Nondescriptives, one hundred generative stills made for a Bright Moments exhibition in Berlin, training a model on text-to-image outputs to sit between organic forms like tree branches and synthetic ones like city architecture. Its sequel, After Us (After Nondescriptives), turned the same imagery into a sound-accompanied, cyclical animation reflecting on how nature might reclaim cities long after humans are gone.
UnBeautiful (2024) fine-tuned diffusion models on two hundred of her own analog Lomography photographs from her twenties, using the resulting glitches and distortions to push back against how large AI models tend to sexualise and idealise women's bodies.In 2022 she showed Time Stills in “FEMGEN”, a gender-equity exhibition at Art Basel Miami Beach.
Machine-hearted (2023), a 100-iteration long-form drop on Kate Vass Galerie's K011 platform, tests where large AI models succeed and fail at expressing complex human emotion.A related body of work, She was machine-hearted, pairs fine art prints with matching NFTs and gives voice to feelings of shame, pain and inadequacy through short first-person titles such as She Felt Her Body Was Failing Her.,
In Family Album, My Grandmother's Memories (2025) she trained Stable Diffusion models on negatives her grandfather shot in Soviet-era Vilnius, Riga and Trakai in the 1960s and 1970s, reworking them into an homage to her grandmother's memory loss in later life.Her work has been shown at Art Basel Miami Beach, Sotheby's New York, Bitforms New York and Venus Over Manhattan, and she has been named one of the Women in Tech Foundation's Top 10 Women in AI.
