Research papers by Sean Carroll produced as part of doctoral research (2021–2025) on AI, archives, and early computer art. The papers examine how large language models, retrieval systems, and computational methods can support archival interpretation, preservation, and curatorial practice.
Author: Sean Carroll
Writer, producer and curator based in Leicester, UK
Improving Historical Algorithm Recreation Through Systems Thinking: A Study Using Large Language Models and Early Computer Art
ISEA2025 – International Symposium on Electronic Art
Summary:
Tests whether a systems-thinking representation improves LLM reconstructions of historic computer art algorithms. Finds improved capture of intent and relationships, with some trade-offs in mathematical precision.
ISEA2024 – International Symposium on Electronic Art
Summary:
Explores how embeddings and retrieval-augmented generation support conversational access to multimedia archives. Frames archives as interactive systems for knowledge retrieval and creative use.
EVA London – Electronic Visualisation and the Arts
ScienceOpen page: HERE
PDF: HERE
Summary:
Proposes an AI-assisted archival framework that supports exploration and retrieval through dialogue. Argues for capturing interpretation as a first-class part of archival knowledge.
EVA London – Electronic Visualisation and the Arts
ScienceOpen page: HERE
PDF: HERE
Summary:
Documents an AI-assisted exhibition workflow using the Computer Arts Society PAGE archive. Shows how AI can speed analysis while leaving interpretive authority with the curator.
EVA London 2022 – Electronic Visualisation and the Arts
ScienceOpen page: HERE
PDF: HERE
Summary:
Documents the reconstruction of Edmonds’ historic interactive artwork using contemporary networked hardware and web systems. Uses modular rebuilding to preserve behavioural structure.
Cite the conference proceedings page (ScienceOpen for EVA, ISEA Symposium Archives for ISEA) as the publication of record. The PDFs in this repository may be author manuscripts.
Publisher licences may restrict redistribution. If a publisher requests removal, the relevant PDF will be taken down and the README will retain only the external publication link.