Zheng Mahler’s The Twenty-Three Thousand Sexes of Schizophyllum Commune and Other Stories (2026) takes us into realms that have haunted modern thought from the outset—realms in which nature and technology, the world and its representation, art and science, and body and thought intermingle.

Born in the context of the Cold War, the cybernetics of the 1950s was driven by the idea that such complex entanglements could be controlled through computation. Yet, today’s polycrises have made the limits of such ‘instrumental reason’ all too apparent. With the rise of AI, the old cybernetic dream of technocratic mastery of the world is gaining new force. Against this backdrop, we must ask how we might imagine post-cybernetic forms of planetary coexistence that are not based on extractivism, exploitation, and exclusion. What role does art play in bringing such forms into being?

Although it is clear that the aporias of modernity cannot be overcome by the theoretical and technological means of modernity itself,^[1] Romantic ideas seem to resurface in discourses on pluriversal or more-than-human worlds: the desire for a holistic relation to the world, a longing for immersive experiences of nature, and a critique of modern reason in favour of spiritual modes of thought.

This return to Romantic ideas should not be simply dismissed as regressive. As Theodor W. Adorno wrote, ‘So long as progress, deformed by utilitarianism, does violence to the surface of the earth, it will be impossible—in spite of all proof to the contrary—completely to counter the perception that what antedates the trend is in its backwardness better and more humane.’^[2]

The challenge, then, is to reinvent the old idea of ‘natural beauty’ as a critical concept for a world that is increasingly permeated by digital technologies. What might be the parameters of a post-cybernetic aesthetics of natural beauty?

A similar set of questions informs Yuk Hui’s Art and Cosmotechnics (2021) and Zheng Mahler’s The Twenty-Three Thousand Sexes of Schizophyllum Commune and Other Stories.

The point of departure for such an inquiry is a post-humanist understanding of nature as developed by Bruno Latour, Donna Haraway, and a range of ecofeminist thinkers: In this view, nature is not conceived as the pure and innocent Other of technological modernity (a cliché also rejected by the early Romantics). As Anna Tsing compellingly demonstrates in her influential ethnography The Mushroom at the End of the World (2015), organisms, such as matsutake mushrooms, are always entangled in technological, social, and economic processes. Similarly, the mushrooms of Lantau, with whom Zheng Mahler made kin in the course of their artistic research, are as much part of the post-cybernetic condition as Hong Kong's skyscrapers or the data streams circulating through digital networks.

Hui, in addition, emphasises that the predominantly Western critique of technology needs to be pluralised if non-extractivist forms of nature–technology entanglement are to become conceivable in the first place. Technology in general, and AI in particular, are not necessarily opposed to holistic, pluriversal forms of world-making.

Yet, AI must neither be underestimated nor overestimated. As Zheng Mahler demonstrates, existing generative models are unable to adequately grasp the diversity of fungal species, let alone their complex interactions with techno-ecological environments. This is largely because generative AI models are trained on datasets that reflect an obsolete anthropocentric view of the world.

More than that, AI training datasets reproduce the logic of the patriarchal and colonial ‘god trick’ of the nineteenth century: The fantasy of capturing and representing the entire world through a rigid taxonomic grid.^[3] It is in this renewed faith in calculability and representation where true—and dangerous—regression lies.

Apart from the fact that every categorical grid is imbued with relations of power, AI fails to account for fundamental dimensions of human experience that are crucial to transforming our relationship to technology and nature. Drawing on Martin Heidegger’s notion of Dasein, Maurice Merleau-Ponty shows in Phenomenology of Perception (1962) that an embodied ‘being-in-the-world’ cannot be reduced to acts of recognition and classification. Instead, it is grounded in an implicit—that is, not fully codifiable—form of bodily knowledge. This knowledge is doubly situated: First, it develops through repeated interaction with the existing socio-material world; second, it is mobilised in response to the concrete demands of a situation. An embodied experience of the world thus resembles the cybernetic feedback loop insofar as it is genuinely recursive, yet it cannot be translated into unambiguous formulas.^[4]

If even the workings of human perception cannot be adequately reproduced by AI (or other technical apparatuses), what, then, of the multiple forms of knowledge and interaction exhibited by fungi?^[5] Any attempt to generate a ‘more complete’ representation of the more-than-human world is necessarily doomed to fail. No matter how sophisticated AI becomes, it remains dependent on formalisation and computation and, ultimately, on digital code. The ‘thing-in-itself’ is—and fortunately remains—an enigma.^[6]

This brings us back to ‘natural beauty’ and the crucial question of form. Due to their probabilistic logic, AI-generated images gravitate towards the most likely. However polished their surfaces, they converge on a narrow, highly biased repertoire of readily identifiable shapes and tropes. Form is not so much reduced as rendered predictable.

AI is therefore incapable of generating ‘aesthetic forms’ in a strictly philosophical sense. Following Adorno, aesthetic form is the opposite of the expected. It consists of extracting the existing materials, signs, and forms of society from their usual contexts and relating them to one another so that, within their constellation, the ‘non-identical’ can appear—that which cannot be subsumed under the established concepts, forms of communication, and cultural techniques of a given historical moment.^[7]

Analogously, Adorno’s concept of ‘natural beauty’ does not refer to what can be consumed as a ‘beautiful landscape’. Instead, it is the appearance of something that remains perceptually and conceptually out of reach. An artistic engagement with the techno-ecological conditions of a given historical moment therefore cannot aim to represent the environment in a naturalistic manner. Rather, through the transformation and reconfiguration of historically available artistic means, it would have to allow precisely that moment of nature to appear, which eludes the grasp of ‘instrumental reason’.^[8]

It is at this point that Hui also marks the epistemological limits of AI: ‘In art and philosophy, the unknown is the object of intelligence and also the condition for the development of such intelligence. This is a significant demarcation between an intelligent machine performing calculative tasks for human agents and another type of intelligence taking the unknown as its subject.’^[9] One key task of art, accordingly, would be to transform technology itself. A post-cybernetic aesthetics of nature thus confronts the paradoxical challenge of making the non-identical, the unavailable, or the unknown appear by means of AI.

This, however, is only possible through deliberate intervention: for instance, by developing alternative training datasets that incorporate insights from (post-)phenomenology, New Materialism, and cosmotechnics. On this basis, AI might perhaps learn to generate images in an aesthetic sense—that is, images that break with naive representationalism and instead open up poetic associations, interstices, and empty spaces.

Zheng Mahler’s custom-made AI aims precisely for such a neo-Romantic fusion of science, non-human agency, technology, and poetry without seeking to fix these ephemeral relations into identifiable forms.