Students Used AI to Score 96% on their exams How Would Your Organization Score on its use of AI?

Students Used AI to Score 96% on their exams How Would Your Organization Score on its use of AI?

Stop Blaming Shadow AI. Start Harvesting It.

A professor at Brown University sent his midterm home. Nearly every student scored 96% or above. He called it cheating.

What actually happened

A Professor at Brown University, who has taught the course for nearly two decades, said enrollment for the semester was unusually high, jumping from a typical range of 8 to 30 students up to 86. Dozens of those students went on to earn perfect or near-perfect scores on the take-home midterm. Serrano suspected AI immediately, and when he moved the final exam back into the classroom, the results told their own story: eighteen students dropped the class, nine stayed enrolled but skipped the final entirely, and the average score on the in-person exam came in at 48.6%, the lowest in the course's history.

(Read the original story: Brown Professor Suspects Most of His Class Used AI to Cheat — Inside Higher Ed)

As Brian put it on this week's episode:

"So we have an approach where maybe the students aren't the failure here... they went and used AI, they must have validated the responses form AI to make sure that they were accurate to get to that grade."

Does this Sound familiar?

It should, because it's happening in your organization right now. Someone in marketing is using an AI assistant nobody sanctioned. Someone in sales is drafting proposals with a tool nobody approved. And when their numbers start looking "too good," the instinct in most companies is the same one that professor had: assume something's wrong, shut it down, punish the people who found the edge.

Campbell framed the deeper issue well:

"It's basically institutional mechanism design... designing a game that practically is begging to be exploited, and then blames the players."

Where the real failure sits

If an employee improves their output using AI, and there was no clear policy telling them they couldn't, the problem was never the employee. The problem is that leadership hasn't done its job, deciding what's acceptable, educating people on how to use these tools well, and building the governance to support it instead of discovering it after the fact.

The move

Don't hunt down shadow AI and shut it off. Find it. Understand why it's working. Turn it into sanctioned practice, with the policy, training, and oversight that should have existed from day one.

Mike said it best closing out the episode:

"It is not a crisis, it is an opportunity."

The Shadow AI exam, is already happened in your business. The only question is whether you're going to grade it as cheating, or harvest it as value.