As an assistant professor of artificial intelligence in a business school, I am determined that my students learn to use AI for everything they meaningfully can. But that creates an obvious problem. If students can use AI to help create the essay, report, spreadsheet, or presentation, how do I assess what they know? More importantly, how do I determine what in their work is uniquely human?
I needed an approach that did not restrict AI use, but instead changed how the work was evaluated. I call it the Three Humans Framework.
In my classes, one of three humans will grade your work. The first is me, acting not simply as a professor, but often as a stand-in for a future boss or manager. The second is a stakeholder from an outside organization, someone with a genuine interest in the problem. The third is another student, someone working alongside you who must understand your reasoning, challenge your thinking, and collaborate with you as part of a team.
When an essay, report, spreadsheet, or presentation can be substantially or even completely generated by AI, the more revealing assessment is how you justify your artifact. Can you explain why you made the choices you made? Can you defend the assumptions? Can you recognize weaknesses? Can you respond when someone challenges your conclusion? Can you change your mind when the evidence warrants it?
That oral defense, whether it happens with a professor, a stakeholder, or a teammate, may become one of the most important ways we determine whether someone truly understands the work in front of them. It also helps answer a question that is going to become increasingly important: Is this AI-generated work, or is this human work created with AI?
The distinction is not whether AI touched the artifact. It almost certainly did. The distinction is whether a human exercised judgment over it.
I recognize that this approach will be uncomfortable for some students. Oral defense pushes people toward things many naturally fear: public speaking, disagreement, difficult questions, and the possibility of being wrong in front of someone else. But that discomfort is not unique to school. It is the basis of good work.
In nearly every meaningful career, you eventually have to explain what you built, defend what you recommended, persuade someone to act, or respond to criticism. And as AI becomes better at producing the work itself, those moments will matter even more. Because at some point, another human will look across the table and ask you to explain yourself. That is the test.
