Description

The TRAI project rests on the assumption that the ongoing AI revolution will have an irreversible impact on academic working and thinking practices. Students are already widely adopting LLMs such as ChatGPT even though consistent teaching offers and institutional usage policies are still in the making. The project aims to understand, analyse, describe, and extract relevant education-related information regarding the relationships within the triad of academic writing, critical thinking, and AI literacy. 

The project addresses this topic in the context of undergraduate thesis writing, a crucial academic practice in most national higher educational systems. Thesis writing is a critical event in the 3-4 year Bachelor’s degree programmes, forcing students, many of them for the first time, to undertake independent, research-based work. From previous studies, we know that this is the point where AI is changing traditional academic routines most profoundly. The TRAI project uses a multidimensional, interdisciplinary methodology to get insight into such changes and involves several interrelated data collection stages. First, a large-scale survey will be conducted to gather valuable information about academic writing with AI from students and faculty. Semi-structured Bachelor’s student interviews will then be conducted based on the survey results (also reflecting potential differences among participating countries). The interviews will be designed to capture the practices and routines of thesis writing and the challenges of AI on critical thinking and epistemic development. Supervisors from the same institutes will be interviewed to understand institutional policies and the teachers’ perspectives. 

A multilingual thesis corpus of Bachelor’s theses (TEZEC) will be created in parallel. The corpus will be collected in two stages, one before a didactic intervention in the form of a massive open online course (MOOC) on AI use in thesis writing and the second after the didactic intervention. One of the core research units in the TRAI project is to build a methodological framework that includes identifying and measuring the linguistic features that are indicators of critical thinking and epistemic awareness in the development of theses. This is particularly challenging in the context of the mentioned triad since it is becoming increasingly difficult to integrate writing elements which are AI-generated with personal ones. Describing and, if possible, automatising the identification of features specific to personal text production, in connection with the inherent critical thought and AI production, is an important dimension of the analysis. 

By triangulating the results of the surveys, interviews and the corpus, insights will be collected on how knowledge is constructed through writing (e.g., assertiveness, authorial voice, linguistic complexity), and thoughts critically evaluated (e.g., argumentation, referencing, analytical thinking, hedges). The project’s ultimate goal is to conceptualise how digital scholarship evolves under growing AI influence, thus providing insights into new forms of intellectual inquiry and institutional responses in higher education. The didactic outcome of the TRAI project, which uses information from all types of data collected (surveys, interviews, and corpus data), takes the form of a multimodal MOOC for students and teachers. Another essential and sustainable deliverable in the project, which incorporates all significant research results, data (i.e., corpus) and didactic materials, including the MOOC, is a multilingual platform (eTEZEC). The platform’s content will be made freely available in all three languages of the project consortium (German, Bulgarian, and Rumanian), plus English for broader impact and visibility of the findings.