Thinking about scientific authorship in the age of large language models: Implications for geographers

Staddon C.; La Matta Romero F.; Adu-Sarfo G.; Hassanin A. · Journal of the Bulgarian Geographical Society (2026) · DOI: 10.3897/jbgs.e188669

Resumen

This paper examines whether large language models (LLMs) can qualify as scientific co-authors, situating the debate within broader epistemological and philosophical traditions. Distinguishing between “instrumentalist” and “purist” perspectives, the authors analyse empirical evidence on artificial intelligence (AI) use alongside critiques concerning opacity, accountability, and epistemic authority. Drawing on Plato’s concept of the pharmakon and Derrida’s theory of textual mediation, the paper argues that while AI increasingly participates in knowledge production, authorship entails normative responsibility and answerability that AI systems cannot bear. The authors conclude with practical principles for governing AI use in scientific writing, emphasising collective epistemic stewardship over technological determinism. © Chad Staddon et al.

Instituciones: Universidad Peruana Cayetano Heredia