Emi Blakely

At the beginning of the academic year, Harvard College dean David J. Deming sent a school-wide email proposing a policy that encouraged professors across all departments to use AI, including in writing courses. Inevitably, this led to immediate backlash from faculty and students alike across the country, indicating significant hesitance toward the drastic changes Deming proposed. I was shocked to hear that one of the top universities in the world would encourage its students to use this tool, whose negative effects are such a salient topic of discussion. However, after reading Deming’s email and thinking more about the subject, I have come to agree with him that this is the best way forward for all schools, including Bowdoin.

As a freshman who has written only one essay in college, I am already keenly aware of how badly the status quo is failing. Students use AI across disciplines and in many different ways, but I’ve observed one constant: homework (and often even in-class assignments) is rarely completed by the student alone. I often find myself impatient when struggling through a problem set and reach for resources that can teach me the material. While I will read through what the AI spits out and use it to improve my understanding, I also realize that I am skipping the struggle, a critical part of learning.

Perhaps even more worrisome than the lost practice of struggle and critical thought is the lack of guidance students receive on how to navigate this technology. Bowdoin students have easy access to this sort of advice: through the comprehensive AI class offerings, AI-related clubs and Hastings Initiative opportunities. I think these still fail to meet the moment. Despite the easy access, the average Bowdoin student (including 69 percent of first years who responded to the Orient’s First Year Student Survey) interacts with AI consistently and receives limited professional input on how to optimize its use. Personally, I’ve heard nothing about AI in my classes beyond the typical syllabus introduction discussion. Bowdoin often boasts about the astronomically large sums of money it has devoted to AI, but the typical class still has not reaped the benefits.

The strongest argument against Deming’s proposal is that it lets students skip the very skills college is supposed to build. I take that concern seriously, but it rests on an increasingly unrealistic understanding of the academic environment. Students are already using AI to bypass precisely these skills, despite policies prohibiting them from doing so. The question, then, is no longer whether students will use AI, but whether colleges will structure that use around learning. Maintaining rules that students can easily circumvent and schools cannot reliably enforce has limited value. Instead, work that requires independent thought can still occur without AI, but it must be done in a controlled environment.

Others in opposition to Deming’s proposal are fundamentally opposed to AI use beyond academia. It is quite rational to be scared of the horizon when it comes to AI—leaders in the field themselves warn us of the perils. We must be aware of the detrimental environmental impacts of AI. Despite the negatives, this technology is clearly part of the future regardless of the evident possibility of extreme consequences. Students should be well-positioned to be part of this changing world.

Bowdoin faculty must act quickly to develop a policy that prevents improper AI use on essays, even if this means drastically shifting how assignments are completed. For example, it is reasonable to conceive of a world where students write first drafts of essays during a class period equivalent to a lab, then AI is used to revise and refine the original piece with the help of the professor. This forces the student to form their own arguments, compose their own prose and structure their own essays before using AI, allowing students to develop their thinking skills as well as continue to grow and adapt with the ever-changing world through their later controlled and sanctioned AI use. Additionally, demanding that students more routinely focus for extended periods of time will help combat ever-decreasing attention spans resulting from short-form social media content, but that’s a different conversation for a different article.

Deming’s email drove the conversation, but it has yet to be a driver of change. Bowdoin should understand that students are not trustworthy in their AI use, and detection tools are ineffective. Thus, the College should stop relying on those two assumptions to enforce rules critical to our learning. Moving the first draft into the classroom is a policy that works without relying on either.

Kaelen Annes is a member of the Class of 2030.