Evaluative conversations as an AI-resilient, effective and inclusive form of assessment

The rise in artificial intelligence has accelerated the need for Higher Education Institutions worldwide to rethink their approach to student assessment. One candidate receiving increased attention in tackling widespread AI usage and to promote academic integrity is the use of evaluative conversations as a form of interactive oral assessment. However, despite qualitative studies suggesting beneficial evidence, to date quantitative analysis of this form of assessment has been lacking. Here, using data from a first-year cohort of undergraduate students we critically assess the effectiveness of evaluative conversations as a form of assessment overall and across two different demographic groups; those with/without declared disabilities and age at start of course (mature vs non-mature). We find evidence that evaluative conversations are an effective and inclusive form of assessment, with students achieving marks similar to online course tests, regardless of demographic group. With an appropriate design in place, evaluative conversations can be a powerful addition to the assessment toolkit of Higher Education institutions.