Verbal defenses effectively detect AI-generated cheating

What's this about?

People disagree about whether schools can spot AI-made cheating by asking students to explain their work aloud.

A talk can help teachers learn more, but it may not prove who used AI.

What supporters say

• Teachers can ask how students chose facts, used sources, and built their main point.
• Sudden, detailed questions may show when a student does not understand their own work.
• A spoken defense makes it harder to hand in AI text without knowing what it says.
• Clear questions, trained staff, and fair score guides can make oral tests work better.

What critics say

• Few studies show that spoken defenses can spot AI cheating in a sure, fair way.
• Oral tests can check learning, but they do not clearly show who used AI.
• A shy, nervous, or less fluent student may struggle, even when they wrote the work.
• A student who used AI may still study the text and give strong answers.

The bottom line

Spoken defenses can give teachers useful clues when they worry about an assignment.

But we do not yet have enough proof that defenses alone can reliably spot AI-made cheating.