Webinar: What student prompts reveal about their learning: Introducing the DRIVE assessment framework for AI-assisted writing

Summary:

Many higher education teachers are facing the need to rethink their assessment practices after the rise of Generative AI chatbots. These tools can easily generate humanlike written assignments. How can then teachers assess genuine learning when students can use GenAI tools for writing assignments? Grading essays in the age of GenAI has turned into a sort of "Turing test grading" task where it is no longer clear whether the evaluation reflects a student's ability to prompt GenAI or if it taps into their learning of course-specific knowledge.

What if we could instead assess student learning by examining their interactions with GenAI?

In this webinar I will discuss the findings of our TU/e research exploring this possibility where the assessment focus shifts from output (essay) to process (interaction with GenAI). Our findings are grounded in authentic classroom data from high-stakes writing assignments, where students could opt to use GenAI, and their GenAI usage was evaluated alongside the essay itself. In this context, we propose the DRIVE framework to guide evaluation of student-GenAI interactions. Our analyses revealed which interaction types were linked with high-scoring essays under traditional methods (output focus), and which demonstrated student learning according to DRIVE (process focus). We will discuss practical implications and explore adapting this approach beyond argumentative essays to other academic writing contexts.