My class has an AI student

On Tuesday, I will sit down in class a few seats away from the newest addition to the Princeton student body: an artificial intelligence chatbot.

According to the syllabus of my entrepreneurship course, the AI “student” will read our assigned case studies and supplemental readings, listen continuously to our classroom discussions, accumulate knowledge throughout the semester, and contribute when called upon by the professor. Conceived by a former student and still under development, hosted on a computer that the TA monitors, it can even summarize what we discussed at the end of class.

I’m curious about my AI classmate. I’m also uneasy.

Some of my uneasiness may come from describing it as a student, or the fact that the course ascribes it a name and pronouns to make it seem more human. But the uneasiness also has to do with what its presence implies. Princeton professors and departments have spent considerable effort regulating when and how students should use artificial intelligence, though those regulations remain inconsistent between courses — and within them. 

In this course, we’re not allowed to use AI to prepare for classroom discussions or presentations. But the “student’s” presence signals a disconcerting position on AI: it might have a physical place in the classroom after all.

The distinction is that this AI “student” is not there to do our work for us. It is there to help us understand what AI can and cannot contribute to human reasoning, which may be a worthwhile experiment. AI is assuming greater significance in the realm of entrepreneurship, and it can be insightful, based on the “student’s” contributions, to understand where or where not we can lean on AI, and how we can work in tandem to make better business decisions. But universities should be careful not to confuse the ability to generate an answer with the experience of discovering why a question matters or the creativity and critical thinking humans use to get ourselves to an answer.

The knee-jerk response to AI in education is prohibition: banning the use of AI on homework assignments, restricting the use of devices during class time, and introducing proctors to exams all aim to ensure students complete work without technological assistance. Those measures can, in some cases, achieve that specific goal, but they do not address the harder question: what do we actually want students to learn, and what kinds of experiences produce that learning?

AI has rendered that question unavoidable, yet challenging to elucidate. If a machine can synthesize information from a case or discussion and generate a relevant, coherent response almost instantly, then an assignment designed primarily to test those abilities may no longer be worth what we thought it once was. Our assignments need to evolve beyond only requiring the skills that AI is a suitable replacement for. In that sense, the “student’s” participation in my class may help us go beyond the perfunctory — because if we can’t produce a better answer than a robot, what sets human intellect apart? 

There is, however, another part of the “student’s” presence that demands attention: it is learning from us. For those who have chosen to take the course, we are contributing to the learning of an AI model, whether we like it or not. At the same time, AI researchers are sounding the alarm about the technology accelerating us to our own doom.

No matter how you train it, though, an AI chatbot will not acquire the background and life experience that human students introduce into a discussion. It cannot recall a childhood memory or conversation that helps make an argument click. It does not have the years of effort working towards a club or internship task that has molded how we now approach a business case study. Frankly, it has not earned the right to learn from us and our experiences.

Perhaps these distinctions point toward what education should preserve.

The goal of education should not solely be to make students better at doing everything AI can already do — reading, summarizing, iterating, designing. It should be to give students experiences that make them better thinkers in ways that cannot be reduced to producing the correct answer.

Some of the most valuable intellectual experiences cannot be easily graded. They happen when a professor says something that stays with you after class, when a friend challenges an assumption you didn’t realize you held, or when you walk into a discussion knowing almost nothing about a subject and leave with a question that you can’t stop thinking about.

So, instead of feeling threatened by the encroachment of AI into spaces dedicated to honoring the highest forms of our humanity — our intellect, ingenuity, and moral reasoning — perhaps it should serve as a catalyst for us to recommit to these human qualities, starting with these academic spaces. 

I don’t know what the “student” will contribute to my class this semester. It might frustrate me, it might surprise me, and, who knows, it might offer insights that I can learn from.

But I hope that by the end, the most important thing we have learned is not what AI can do, but what remains uniquely ours to do.

Emily Zhang ’29 is among ORFE majors wary of the expansion of AI within her major and across the University generally. She is a sophomore Opinion columnist from Poughkeepsie, N.Y., and can be reached at emilyzhang[at]dailyprincetonian.com.

Editor’s note: As a courtesy to the instructor of the ongoing course at hand, and to avoid disrupting the instructional process, The Daily Princetonian is not naming the course in question in this piece.