Time for the ‘wiki-thesis’: Rethinking research assessments in the age of AI writing

By Salil Sahadevan & Deepak John Mathew

AI writing tools, capable of generating polished prose and synthesising complex information, nudge us to reconsider how we assess student learning. Artificial intelligence enhances productivity in the writing process. It also poses questions on how we use it and where it leads to.

Recent research by Khalifa and Albadawy (2024) note that AI can assist a research student at least in six areas – idea generation, content structuring, literature synthesis, data management, editing, and ethical compliance.

Yet, if a student can produce a seemingly polished thesis without genuinely engaging with the material, how can we ensure academic integrity and meaningful learning? To authentically assess student understanding in an AI-augmented world, we need to re-evaluate the forms that scholarly work can take and the methods used to evaluate it.

Multimodal thesis

Knowledge creation and dissemination need not be limited to the written word.

A multimodal thesis is not confined by conventional text. It incorporates diverse media, such as videos, visualisations, interactive simulations, and audio recordings, to present research findings in engaging ways.

This approach offers a richer expression of research findings and better evaluation of student engagement compared to traditional dissertations.

Students can use multiple data elements and even artistic expressions as evidence for understanding of their research topic. In some form, this is already in existence. There are research students who present interactive maps while exploring migration patterns or documentary films, while showing the impact of climate change on a local community or a simulation allowing users to experience the ethical issues faced by healthcare professionals etc.

Also read UGC to award ‘PhD Excellence Citation’ annually to honour doctoral research

The idea is not just to engage the audience but provide evaluators with a more holistic view of the research process and critical thinking skills the student gained in the research process.

Longitudinal thesis

A process-oriented approach to theses evaluation, like an enhanced version of formative assessment in teaching, might play an interesting role in the search for originality and uniqueness.

This approach values the intellectual development and research journey of the student throughout the research period, not at specific intervals or at the end alone. In this way, a longitudinal thesis is developed over several semesters, allows iterative refinements, continuous feedback, and a transparent view of student learning.

In the self-regulated learning environment of research, it could appear to be metacognition. Evaluators can track the growing knowledge of the student, their interaction with sources, as opposed to AI summaries, and their methodological decisions and biases. Such longitudinal theses can offer a more genuine evaluation of the research progression by providing interim checkpoints and recording every step of the research process, including the moral and responsible use of AI technologies.

The wiki-thesis

Interdisciplinary thesis projects encourage research students to pool their expertise to tackle interlinked problems that demand a multi/cross/trans-disciplinary approach. In an interconnected world, the ability to collaborate effectively, synthesise diverse perspectives, and apply knowledge across disciplines becomes important.

So, longitudinal thesis as interdisciplinary projects is a true work of research as it mirrors the collaborative nature of real-world research and problem-solving. It can trigger better research evaluation too.

A wiki-thesis, built collaboratively on a public platform, can also be a portfolio of student work. This is a possibility which is already emerging on the margins of research institutions.

In this dynamic format, students showcase their individual contributions, revisions, and engagement with peer feedback and show their evolving research and learning process. While not suitable for all disciplines, this approach is relevant in fields that value collaborative knowledge production and iterative development. In any case, it is better than a traditional static thesis document.

Rebellious research and transition worries

To prevent the uncritical use of AI by students and scholars, evaluation can incorporate principles of rebellious research, an idea advocated by Uwe Flick in 2018.

Rebellious research emphasises on research that accommodates diversity and contextual relevance. By encouraging students to critically examine their use of AI tools, to engage with diverse perspectives, and to consider the ethical implications of their research, a more responsible approach to knowledge creation can take place. Most institutions are yet to make the use of AI writing tools a part of research coursework.

Also read Agriculture Courses: ICAR makes internships a must, adds skills training in revamped curriculum

The transition to these new forms of theses raises valid concerns about fairness and consistency in evaluation. How can we ensure that students are assessed on a level playing field when their projects take such diverse forms?

Clear rubrics, increased number of data points for evaluation, and a progressive and alternative vision of research are the answers. Such transition worries pale in comparison to the urgent demand for ethical and transparent AI uses, and evolving use of originality detection tools and humanisation tools. In spite of any tools, the future of scholarship depends on our ability to adapt and uphold the values of critical thinking, originality, and intellectual rigour.

Deepak John Mathew is professor with IIT Hyderabad and Salil Sahadevan is Deputy Secretary with University Grants Commission (UGC). Views are personal.