Do indicial ao algorítmico: promptografia e a era das máquinas de imagens generativas IA na construção de mundos possíveis

Abstract

This thesis investigates how generative image machines, based on Artificial Intelligence (AI), can create possible worlds that go beyond the traditional indexicality of photography, by analyzing the epistemological implications of these new regimes of verisimilitude. Starting from the assumption that photography has historically been linked to a physical referent that guaranteed its status of truth, the study examines how algorithmic simulation reconfigures the relationship between the real, the fabricated, and the imaginary, establishing new forms of visual credibility. The research seeks to understand not only the technical fundamentals — algorithms, machine learning architectures, and databases — that support these systems, but also how human agency is transformed, migrating from the photographic act to the operation of prompts and the curation of results. The study characterizes the ontology of generative images, contrasting their “engendering” process with the indexical paradigm of photography, and evaluates how the absence of a direct connection to a physical referent challenges notions of authenticity and truth. By articulating technical analysis and conceptual reflection, the research investigates how the algorithmic production of images expands the limits of human imagination, generating unprecedented visual repertoires that blend aesthetic codes and cultural references. By discussing the epistemological implications of the AI-produced imagefiction, the thesis proposes that these new possible worlds constitute not only a technological phenomenon but also a cultural transformation that demands the revision of theoretical and critical paradigms about image, verisimilitude, and visual knowledge in the contemporary context. Finally, it is argued that understanding this new ecology of images requires an interdisciplinary perspective, capable of integrating image theory, philosophy of technology, and data science. By recognizing that contemporary visuality is built at the intersection of human agency and algorithmic operation, this research seeks to offer theoretical and critical subsidies to think about the role of the generative image in the construction of visual artifacts. By analyzing our corpus, it is concluded that generative images not only expand the modes of imagining but also establish regimes of visuality in which the creation of possible worlds is converted into a collective, poetic, and epistemic practice, pushing the boundaries between real and fictional.