Artificial Intelligence as a Learning Tool for Identifying and Categorising Artworks from Computational Creative Practice and Historical Computer Generated Metadata
Sean Leslie Carroll PhD Thesis, De Montfort University, September 2025 Faculty of Art, Design and Humanities (ADH) Funded by Midlands 4 Cities Doctoral Training Partnership
This thesis investigates the role of generative AI in the preservation and interpretation of computer and digital art. These works are fragile, often surviving only in partial records or unstable formats, and archives struggle to represent their originality and context.
The research asks how AI can preserve and re-activate digital artworks when viewed through a constructivist and genealogical lens. Three practice-led projects form the centre of the study:
- New Generations -- triangulated vocabularies of practice through archival encounters
- Re:Constructs -- operationalised systems rubrics in AI-mediated reconstruction
- Listen, Scoundrels! -- activated discursive traces through conversational systems
The thesis contributes three key concepts: parallel provenance (a genealogical model for representing originality), epistemic legibility (a principle for preservation), and interpretive infrastructure (a framing for AI's role in heritage).
This repository contains the PhD thesis document (without appendices).