I write about leadership, education, AI, and why teaching critical thinking about it is more important than ever.
Please hit the heart ❤️, restack 🔄, subscribe 📨, and all that jazz to help spread the word!
🙌
Well, it finally happened. A university professor decided enough was enough.
They were tired of grading perfectly written papers that sounded like they’d come straight out of a McKinsey slide deck with flawless output.
The arguments, so coherently and fully articulated, would make a lawyer proud. But yet, when students were asked to explain their own work and thoughts, they couldn’t.
This was at NYU’s Stern School of Business, where a professor started noticing that student submissions were reading more like professional consulting memos than student written essays. And when he asked them to explain what they had written, well, that’s when the AI rubber didn’t quite meet the road.
So what did he do? He stopped arguing about it and did something that’s going to shock most educators.
He brought back oral exams. Now, I know what you’re thinking. That’s not really news because many faculty are doing this. But wait for it…
The most fascinating part? He used AI to do it.
He didn’t lecture the students, and it wasn’t some old school return to Kelly Blue Books. He worked with AI and came up with a way to determine whether students actually understood what was being taught in class.
And that’s when he introduced AI-powered oral exams.
He built an AI interviewer using ElevenLabs speech technology that asked students questions out loud, followed up on their answers in real time, and pressed them to explain their reasoning. All without the students being able to access notes or ChatGPT for help.
Now, of course they were being monitored. But they were asked to literally talk through their work, conclusions, and how they got there.
Suddenly, they not only had to understand the memo but also explain it—on their own.
Over nine days, the system examined 36 students. Each session ran about 25 minutes with the total cost coming out to roughly $15 per student.
That number alone is going to make a lot of people uncomfortable.
This one scaled in a way that was both efficient and cost-conscious without sacrificing rigor.
There was a human still involved, but their role was logistical. They made sure everything ran smoothly and that the students were staying within the testing guidelines.
But here’s where it gets controversial for some.
After the oral exams were completed, the professor didn’t grade them himself. He fed the transcripts to what he called a council of AI graders. Claude, Gemini, and ChatGPT all reviewed the same material and scored it independently.
According to the professor, AI did all the scoring, but he supervised, validated, and interpreted the results. He believed this was more fair and less prone to human subjectivity.
He set up a rubric and let the AI make the individual assessments. Now, of course, he participated in reviewing and made the final decision.
Students, for their part, did not describe the experience as relaxing. But oral exams aren’t really supposed to be. Several said it was stressful. Some found it intimidating. But many also admitted that it rewarded real understanding in a way written assignments no longer reliably do.
You can’t hide behind formatting when someone asks you to explain your choices out loud.
We have reached a point where written work or anything assigned outside the classroom is no longer a reliable way to measure comprehension. Regardless of the reasons, the outcome is that the technology has changed how easy it is to generate competent-looking output.
And notice what this professor didn’t do. He didn’t twist himself into a pretzel trying to change what was being evaluated, or ask students to pretend AI didn’t exist. He came up with a solution using the technology.
He built an evaluation system that made it immediately apparent who knew the content and who didn’t.
It really does make you wonder if this is where we’re heading with other types of assessment in higher ed?
Now let’s look at what the professor called this, “fighting fire with fire.” That’s an interesting phrase to use.
It does seem to capture how many people experience this tech, especially educators. It can feel like you’re at war constantly, trying to figure out how to reach students and verify knowledge.
And maybe the answer isn’t dying on the hill of purity or pretending AI doesn’t exist. Maybe it’s integrating the technology and responding with updated ways of assessing mastery just like this professor did.
If this were just another breathless story about AI grading students or using injections to catch cheating, it would be easy to dismiss. But this is something else entirely.
This is the culmination of two forces colliding:
Students using AI proving that written work is no longer proof of thinking
And a professor outthinking the AI by joining forces with it, using it to bring reasoning and learning back into the classroom.
Ironically, AI didn’t kill the oral exam.
But it may have just brought it back.