Between Automation and Authorship: Generative AI and the Doing of Academic Writing
Short description
A four year research project (2026–2029), funded by the Swedish Research Council (Vetenskapsrådet) and led by Professor Thomas Hillman. The project investigates how Generative Artificial Intelligence (GAI), particularly large language models (LLMs), is reshaping higher education, with a focus on academic writing and students’ work with master’s theses. Rather than approaching AI in academic writing primarily as a question of academic misconduct or the need to develop new AI literacies, the project examines GAI as an epistemic technology that transforms how knowledge is produced, how authorship is understood, and how academic work is assessed.
Drawing on digital traces of writing processes, interaction analysis of prompt-and-response logs, a series of interviews, and, policy analysis, the project investigates how GAI is used in academic contexts. Grounded in socio‑cultural perspectives on learning and writing, the project aims to develop a nuanced understanding of how AI reshapes academic work and institutional practices in order to inform responsible, evidence‑based approaches to integrating these technologies in higher education.
Background
Generative AI (GAI) has quickly become an essential part of everyday life in higher education. Since tools like ChatGPT, Copilot and Claude became widely available, students and teachers have had to navigate new ways of producing, evaluating, and discussing academic writing. Because writing is one of the most important ways students learn to think, argue, and participate in academic communities, the arrival of new forms of AI raises fundamental questions about how knowledge is created and how authorship is understood. Universities are now facing rapidly changing expectations, new writing practices that are sometimes invisible to educators, and uncertainties about how to support learning under new conditions.
Aim
The project aims to understand how GAI is reshaping academic writing and the epistemic practices that surround the writing of texts, through an investigation of how students, teachers, and institutions actively adapt their writing practices, teaching approaches, and assessment methods when AI becomes part of the writing process. The goal is to develop a nuanced understanding of how AI becomes integrated into everyday academic work and to provide insights that support responsible and pedagogically sound use of AI in higher education.
Research questions
The project explores three central questions to explore how universities respond to AI, how students use AI in their writing and how supervision and assessment are changing:
- How do institutional policies and guidelines define the role of AI in academic writing, ethics, and assessment, and what assumptions about knowledge and authorship shape these responses?
- How do students incorporate AI tools into their thesis writing, how does this influence their ways of working with knowledge, and how do they judge the reliability and quality of AI‑generated text?
- How do supervisors and examiners understand and evaluate student work when AI is involved, and how do they ensure that academic standards, disciplinary knowledge, and critical thinking remain central?
Methods
The project combines several complementary research methods to explore how generative AI is used in academic writing. Policy analysis examines how universities formulate rules and expectations around AI, while interviews with supervisors and examiners shed light on how teaching and assessment practices are evolving. Writing workshops allow students to freely experiment with different uses of AI, and give researchers access to the prompt‑and‑response interactions that unfold during this exploration. Student interviews use a trace‑interview approach, where students walk researchers through their writing processes and AI interactions. A specially developed visualization tool shows when and how students use AI during their thesis work, helping researchers discuss key moments in the writing process. Together, these methods make it possible to study AI’s impact on academic writing from institutional, pedagogical, and epistemic perspectives.
Participants
Thomas Hillman
Charlott Sellberg
Ann-Marie Eriksson
