QuizBuddy

Written take-home work stopped being reliable evidence of individual understanding the moment generative AI could produce a competent answer in seconds. Oral examination is the obvious alternative, and it does not scale: thirty students at ten minutes each is five hours per assessment.

If the constraint is instructor time rather than pedagogy, then automate the conversation, not the judgment. QuizBuddy conducts a structured spoken exchange with each student and returns a transcript and evidence summary the instructor reviews.

• Asks a student to explain a concept, result, or decision in their own words.
• Generates adaptive follow-up questions based on what the student actually said.
• Produces a transcript plus a summary of what the student demonstrated and where they hesitated.
• Hands judgment back to the instructor rather than issuing a grade autonomously.

Sits squarely in Academic Integrity — verification of individual understanding — and in Student Learning, because explaining aloud is itself a retrieval and consolidation exercise.

• Evidence, not verdicts: the tool never assigns a final grade on its own.
• Low stakes, high frequency beats one high-stakes oral exam.
• Transcripts are visible to the student, so the process is legible rather than opaque.

Immediate feedback and the mere act of committing to a spoken answer raise the salience of gaps in understanding. Frequent short interactions also reduce the anxiety premium attached to a single decisive exam.

Piloted in graduate and undergraduate analytics courses as a complement to written work, not a replacement for it.