ChatGPT and the Cost of Convenience

Harvard recently opened its arms wider to ChatGPT.

This school year, Harvard students, faculty, and staff will have access to OpenAI’s ChatGPT Edu. Tailored for higher education use, ChatGPT Edu was built for universities to “responsibly” bestow their inhabitants with the magic of AI. As a Writing Center tutor who works with my peers every week to learn by struggling through the difficulty, I believe this tacit endorsement of ChatGPT puts Harvard’s values as an institution in peril.

Aversion to the growing use of ChatGPT in academic spaces is nothing new. This aversion stems from fears of technology becoming a substitute for critical thinking skills. While I’m wary of outright rejecting the inevitable development of AI, I believe GPT Edu jeopardizes such inquiry and is incompatible with Harvard’s motives of fostering intellectual collaboration.

Our first steps through Harvard’s gates are, we’re often told, the beginning of a mythical “transformative experience.” This experience is grounded in the intellectual and social growth offered by each interaction on our campus. Per the University’s mission statement, these interactions can be found not just in the classroom but everywhere — the dorm, the dining hall, the library — in our diverse living environment.

I know this firsthand. Much of my own college experience has been defined by late-night study sessions, last-minute group problem set solving, and the occasional coffee chat — when a friend and I actually follow through on the promise to catch up made in passing. It is in these moments of intellectual intimacy — not the hours spent glued to my screen — that I have come closest to realizing the ever-elusive promise of “transformation.”

One could argue that in their rawest form, these intellectual exchanges are based on merely obtaining answers. Collaborative interactions, then, are worthwhile because they’re good at getting us to solutions. By this logic, I might as well get my inquiry answered within seconds by ChatGPT.

But this analysis does not capture the synergistic power of the face-to-face exchange. There is a mutual seeking of understanding involved in conversations, a process of generation. This observation is named by the popular notion of conversational ‘flow’: Our utterances do not just sit in the air — they are challenged, questioned, and expanded by one another. While our intellectual exchanges may be brief, they are anything but isolated or transactional, each its own messy and dynamic act of reciprocity.

As one of the head tutors at the Harvard College Writing Center, I have the privilege of stimulating and experiencing such interactions on a near-daily basis. When students stumble into the basement of the Barker Center in a state of disarray, I do not simply start spitting out solutions like a search engine. Rather, I employ a careful, collaborative approach. The squeaking of Expo markers accompanies sounds of laughter and occasional grumbles of shared puzzlement. Their thoughts — which take the form of scribbles on a whiteboard — lead to more wonderings , until together we fully flesh out the argument that had been lying in wait just under the cognitive surface.

The irks, gut feelings, and rich anecdotes surfaced by bouncing ideas off another human are lost when one succumbs to the undeniable ease of ChatGPT. When not hand-delivered an answer, students work towards an answer — one that chiefly resonates with the student in virtue of their intuitions, knowledge, interactions, and experiences. A personal, unique answer.

In other words, by choosing convenience through ChatGPT, we wrong ourselves. The unresolvable becomes resolvable. The unexplainable is explained. Our own imaginative visions are tainted as we rely on ChatGPT’s homogeneous, recycled output — a pale imitation of human ideation.

Now, I certainly do not expect the average Harvard student to sacrifice convenience for the worthwhile but laborious peer-on-peer intellectual endeavor, though such intellectual collaboration is undoubtedly a fundamental dimension of what it means to attend Harvard. I don’t even oppose the University’s move to provide GPT Edu, for the most part. I do, however, ask the College to do what it takes to make students fully understand the weight of the seemingly inconsequential decision to use ChatGPT to finish a discussion post or cram a problem set.

From ChatGPT’s demonstrable biases (which immediately shatters any illusion of accuracy and objectivity), to the astronomical environmental impact it triggers with a single press of the return key, to missed opportunities for the hard work of learning, students must be fully informed of the costs of AI’s convenience. If these consequences continue to go brushed aside, so too will the value we put on true inquiry and, consequently, the spirit of our institution.

Lauren A. Kirkpatrick ’26, a Crimson Diversity and Inclusivity Chair and Editorial editor, is a Philosophy concentrator in Mather House.