Sofia Crespo — the artist known for using machine learning to imagine speculative organisms — presented new work this year in a practice she has helped define: artificial biology as art. Crespo trains models on the vast visual archive of the natural world, jellyfish, insects, corals, deep-sea creatures, and uses them to generate lifeforms that don’t exist but feel like they might, uncanny and beautiful specimens from a parallel evolution. Projects like her ongoing Neural Zoo treat the neural network as a kind of speculative naturalist, recombining the textures and structures of real life into organisms that never were, and asking what that says about how we see nature in the first place.
Watch: AI-Generated Creatures That Stretch the Boundaries of Imagination — Sofia Crespo (TED) (YouTube)
Nature refracted through the artificial
Crespo’s work sits at a rich intersection: the deep history and diversity of biological form, and the generative, recombinant logic of machine learning. Because her models learn from real natural imagery, their inventions carry the genuine textures and structures of life, which is exactly what makes them so convincing and so unsettling, they look like they belong in a tide pool or a specimen drawer, yet no such creature ever lived. That tension is the point. By showing us plausible life that evolution never produced, she makes the actual astonishing strangeness of real nature newly visible, and she probes how much our sense of “the natural” is itself shaped by images, categories, and now algorithms. Among this site’s AI artists, she’s distinct for pointing the technology outward at biology and wonder rather than at faces, data, or the self.
Artificial life as a way of seeing
Crespo’s project connects to a longer lineage of artificial-life art and to urgent contemporary questions about biodiversity, extinction, and our relationship to the living world. Generating creatures that could have been is a quietly ecological gesture, a way of dwelling on the sheer inventiveness of life and the fragility of the real diversity we’re losing. It also resonates with the biofabrication and living-materials threads this site has covered in research, the sense that biology and technology are increasingly entangled mediums. Her work is proof that AI art’s most interesting territory isn’t always realism or spectacle, but wonder, speculation, and a renewed attention to the world the machine learned from.