N.A.S.T.R.O. Learning: Neural Automatic System [For] Textual Reorganization [And] Output

What happens when the author becomes a writer of instructions and delegates the synthesis of the text to a neural network?
In this experiment, the artificial intelligence neither amplifies nor replaces the human voice, but intercepts its coordinates. The author does not directly modify the generated texts: they intervene through further instructions, guiding the network through a series of iterations toward a final form. The process, not comparable to the well-known experiments of sought/found poetry, is based on a radical methodological fracture and on upstream control over the relationship between form, style, and instruction.
The use of AI highlights two crucial aspects: on one hand, the increase in textual consistency and coherence over time; on the other, the surprising affinity between the generated texts and certain contemporary research-oriented writing. If a network can now simulate these outcomes, does it still make sense to explore these paths?
The included asemic pages offer an additional critical insight: although formally impeccable, they remain devoid of the technical and gestural complexity of human signs, which to this day cannot be imitated.
This project therefore does not merely document a dialogue between human and machine, but interrogates the current boundaries of writing.
In a nutshell, one could say that this book was not written "by" an AI model, but, in every part, "through" a suitably trained and guided AI model.
From a content perspective, the result is cohesive, animated by witty and sometimes surreal ideas, and suffused with a constant theoretical tension-made all the more remarkable by the fact that it arises from the author inviting the AI to reflect on its own generative dynamics and their conceptual implications.