Spinoza distinguished natura naturans, nature as creating force, from natura naturata, nature as created things. Which category fits generative AI? Perhaps both, since its models are built from things already made. But I am more interested in the first. Like a forest, AI generation is a self-organizing system: it does not store pictures but grows them, each image emerging from a latent space the way a branch emerges from a trunk.
In Natura, this generative process is given an appropriate subject. Interiors and forests intertwine. A massive limb crosses a room like a beam. Bookshelves — my recurring motif, the classic architecture of accumulated human knowledge — stand inside a tangle of branches, their contents indistinguishable from nests and roots. Human figures appear among the shelves and branches not as owners of these spaces but as one more species inhabiting them.
The images are rendered in the visual language of fine pen and graphite drawing, the medium through which Europeans first studied and catalogued nature, from Renaissance botanical illustration to nineteenth-century scientific engraving. But here this language of observation is redirected: the AI is not recording nature, it is enacting it. What we see is not landscape as naturata — nature depicted — but a trace of naturans: a machine that, like nature itself, endlessly generates form.