Geospatial phenomenology in literary creativity: a comparative study of human and AI textual constructs

This study aims to understand how the human epistemological sphere differ from artificial intelligence (AI) in producing non-specific texts. The research hypothesis is set to comprehend human written production and how it is deeply rooted in space perception while its related production is oriented and constructed via geographical markers that are totally absent in AI productions. Using a simple prompt as a starter for a broad investigation of the aesthetical output, this paper will examine statistically the main features regarding both sides. By analyzing a dataset of 64 human-authored and 180 AI-generated texts produced in response to identical literary prompts, the research employs a computational framework adapted from media narrative analysis to identify distinct patterns in textual construction and interpretation.
The methodology integrates natural language processing techniques such as document embedding with BERT, topic modeling using Latent Dirichlet Allocation (LDA), and sentiment analysis. To assess phenomenological markers, it has been employed keyword extraction (TF-IDF and RAKE) and semantic graph analysis, identifying human-specific geographical and experiential references within the texts. Statistical methods, including PCA for dimensionality reduction and logistic regression, are applied to compare patterns across the corpora. Visualization tools such as heatmaps and annotated semantic graphs elucidate thematic and structural divergences.
Preliminary findings indicate significant disparities between the two groups. Human-authored texts exhibit rich, contextually grounded depictions of spaces and experiences, often interwoven with cultural and historical nuances. By contrast, AI-generated texts demonstrate repetitive syn-tactic structures and a lack of experiential depth, reflecting their reliance on probabilistic language models rather than embodied cognition. Notably, the absence of geographically specific references and phenomenological engagement in AI texts supports the hypothesis.
This research contributes to broader debates on AI's limitations in creative domains, underscoring the role of human phenomenology in literary production. By interrogating the boundaries of machine creativity, the study highlights the irreplaceable value of human cognitive and cultural frameworks in interpreting and recreating reality through literature.