Support, Not Surveillance: Rethinking AI use for Collegiate Writing
Aubrey Valdez, The City College of New York
Bio
Aubrey Valdez is an Adjunct Lecturer in Psychology at The City College of New York, teaching Infancy and Development and Introduction to Psychology through CUNY's College Now and Early College programs. A queer educator and member of a federally recognized tribal community, she holds an MA in Psychology from CCNY and brings a research background in visual neuroaesthetics to a classroom rooted in authentic engagement and inclusive design.
Course Setting
I teach at The City College of New York, a public CUNY campus in Hamilton Heights with a predominantly BIPOC, first-generation, and working-class student body. My courses — Infancy and Development (PSY 24600) and Introduction to Psychology (PSY 10200) — are offered through CUNY's College Now and Early College programs. My classes are mostly college students, with a small number of high schoolers from diverse cultural and educational backgrounds.
Problem of Practice
In recent writing assignments, I noticed several papers showing signs of AI-generated content: generic phrasing, surface-level analysis, and fabricated citations. I suspected students were turning to AI because they felt overwhelmed, lacked confidence, or didn't realize their authentic ideas mattered more than polished prose. This is an equity issue — "polished" academic prose is racially and class-coded, and students who've been told their voice doesn't belong in academic writing are the most likely to outsource it. My driving question became: How might I increase students' confidence in using their own ideas in college-level writing so they're more likely to complete essays independently rather than relying on AI tools?Strategy
I implemented three interconnected strategies designed to support students rather than police them. First, an AI Use Agreement: a Google Form all students reviewed and signed, outlining unacceptable uses of AI (generating ideas, outlines, drafts, edits, or paraphrasing) and explaining that I value their authentic ideas more than polished prose. Imperfect writing is never penalized. Second, two scheduled Zoom writing support sessions for brainstorming, drafting, and editing, plus individual check-ins with every student about their paper; not just those who self-advocate. Third, authenticity-focused rubrics for both major essays that explicitly deprioritize grammar and eloquence in favor of genuine thinking and engagement with course material. Students often turn to AI when they feel overwhelmed, unsupported, or convinced their voice isn't good enough. Punitive policies disproportionately harm students already marginalized by gatekept academic conventions. By naming expectations clearly, lowering the barrier to support, and signaling through grading that authentic thinking matters more than polish, I aimed to make doing the work feel achievable and gratifying rather than overwhelming.
Documentation
AI Use Agreement
A signed Google Form outlining the course AI policy, what counts as unacceptable use, and the rationale behind it. The agreement reframes academic integrity as a collaborative commitment rather than a punitive policy, emphasizing that authentic ideas matter more than polished prose and that support is readily available.
Authenticity-Focused Essay Rubric
The rubric used for both major essays was designed to explicitly deprioritize grammar and eloquence in favor of genuine thinking and engagement with course material. It signals to students that their authentic voice is what's being evaluated, giving them permission to write like themselves.
Measuring Impact
Post Essay-Writing Process Reflections
Short reflections submitted after each major essay in which students describe how they generated ideas, what steps they took, and where they encountered challenges. Posted as a Brightspace discussion board, these reflections build metacognitive awareness and track whether students are developing stronger, more independent writing habits over time.
Student Reflection Statistics
Student reflections on the strategies applied this semester.
Analysis
My strategies were successful in addressing my problem of practice. AI-flagged papers dropped from six cases last semester to one this semester. That single case was caught because of the discussion board vs. essay baseline I built into the strategy. The student resubmitted authentic work after a direct conversation, which felt like a real outcome rather than a purely disciplinary moment. Midterm essay averages also rose to 97% this semester compared to 91% and 81% in prior sections, and the essays showed human error, idiosyncratic phrasing, and genuine argument, which are visible markers of authentic writing rather than polished AI output.
Studying the writing process reflections shifted how I understood the strategy's impact. I predicted the rubric would permit students to write authentically; what I didn't anticipate was how specific and thorough their reflections would be. Students described using the CCNY library database, building outlines, and iterating on drafts — concrete evidence of independent writing habits forming in real time. Reading these reflections gave me more confidence that the rubric and agreement were doing more than reducing AI use; they were actively building metacognition.
Challenges still remain. The midterm average comparison is partially confounded by cohort differences — prior sections included more high schoolers. Zoom session uptake was also partial, so I don't yet know whether non-attendees felt adequately supported or simply didn't need it. Next time, I'd build a brief mid-semester confidence check-in to capture that data directly rather than inferring it from end-of-semester feedback.
Recommendation
This strategy is well suited to CN and EC contexts, where students are navigating the leap to college level writing under real pressure and are often the most vulnerable to the message that their voice is not appreciated in academic spaces. A few pieces of advice: First, pair every high-stakes writing assignment with a low-stakes authenticity baseline: discussion posts, in-class writing, anything where students' real voices show up regularly. Without it, you have no reference point. Second, be very explicit in your rubric that polish is not the goal; this single shift gave my students visible permission to take risks. Third, talk to every student about their paper, not just the ones who attend office hours, because the students who most need support are often the least likely to ask for it. Frame the policy as support, not surveillance.