Oral assessment won't save higher education from generative AI

Photo by Priscilla Du Preez 🇨🇦 on Unsplash

As generative AI transforms how students produce written work, the knee-jerk response from many institutes of higher education has been to pivot towards oral examinations and presentation-based assessments. It’s an understandable response, because students can’t hide behind ChatGPT during a live conversation (well, until LLM wearables are ready for primetime).

But oral assessment, while relatively AI-proof, can also exacerbate existing pedagogical inequalities.

It’s easy to see the appeal in oral assessment for those who are AI-anxious. It promises authenticity, a direct, unmediated encounter with student knowledge. Oral assessment sets up a direct interaction between the student and the examiner—no copy-and-paste or algorithmic assistance. Just a student, an examiner, and what they both know of a given topic.

But this is a romanticised view that ignores several uncomfortable realities.