Professor Sandra Leaton Gray | One one-and-a-half-hour session | Autumn: 10:00 - 11:30 (UK time) Monday October 20

NB: This course accrues 1 training point.

This 90-minute seminar is designed for early career researchers concerned about the growing influence of generative AI on the integrity of scholarly work. As large language models (LLMs) increasingly shape the digital environments in which research is conducted, foundational practices, such as citation, source verification, and the construction of literature reviews, are becoming more unstable and harder to trust.

The session critically examines how synthetic data and recursive model training risk contaminating research outputs, eroding the traceability of sources and disrupting the historical continuity upon which scholarship depends. Drawing on key theoretical frameworks - Bourdieu on symbolic capital, Foucault on discursive authority, and Latour on inscription - the seminar positions LLMs as active participants in the production and legitimisation of knowledge, rather than as neutral tools.

Alongside this critical framing, the session offers practical strategies for maintaining rigour in AI-saturated environments: building and verifying corpora, documenting provenance, and navigating the ethical challenges of working with automated systems. The emphasis is on equipping researchers with both conceptual clarity and practical tools to respond to rapidly changing research conditions. Suitable for PhD students and early career academics across the humanities and social sciences.