Book – Le fossile artificiel

Author project within the CAS Generative Data Design at the Bern University of the Arts (HKB). This book results from an experiment based on a first draft of the novel The Artificial Fossil. The goal: to transform a literary text into a dataset and then into a printed visualization.

A Python script both split the manuscript sentence by sentence to build the table and, via an API, submitted each sentence to GPT-4, which assigned scores from 0 to 10 to emotional registers (fear, joy, sadness, anger, surprise, disgust) and to thematic categories (science, philosophy, politics, media, artificial intelligence).

This process produced nearly 6,000 entries, each sentence becoming a quantified profile. With p5.js, these data were translated into visual compositions: colors, positions, and intensities following the sequence of the sentences.

The novel here is no longer to be read, but to be observed as a cartography of rhythms, distributions, and gradients.