#14. Composition’s New Ethic of Service: The Politics of AI and Student Voice

This piece is a developed version of ideas I forwarded in previous blog posts about composition's ethic of service and its new iteration with the emergence of AI. Enjoy! 

***

In their contribution to the edited collection Bad Ideas about AI and Writing, Xiao Tan and Paul K. Matsuda (2026) deconstruct a widespread assumption among writing instructors that student usage of AI writing technologies detracts from or erases student voices. Tan and Matsuda argue that instructor expectations surrounding voice do not consider student intention, as students may deliberately use the Standard Academic English (SAE) dialect characteristic of AI writing technologies to satisfy perceived academic genre conventions. The authors also contend that while concerns about voice often center around “cases where the student mindlessly substitutes their writing with machine generated texts,” this focus causes instructors to overlook or dismiss appropriate ways that student writers might collaborate with AI (p. 230). Tan and Matsuda thus recommend that writing instructors help students “become more aware of what voice is appropriate and desirable in different rhetorical situations” and offer instruction for how AI writing technologies can serve to shape and craft such voices (p. 231). Their article raises a larger question related to the overall enterprise of college writing: Who is being served from such conversations about AI and student voice? 

While Tan and Matsuda (2026) highlight more recent writing studies scholarship which argues that voice should be understood as having a polyphonic quality, their article suggests that a traditional, expressivistic understanding of voice has reattained supremacy in the post-AI writing classroom. Such an understanding of voice is originally articulated by seminal figures of the field like Peter Elbow (1973), who states that each student has their own unique voice and that the goal of writing instruction is to develop its potency: “In your natural way of producing words there is a sound, a texture, a rhythm - a voice - which is the main source of power in your writing. I don't know how it works, but this voice is the force that will make a reader listen to you, the energy that drives the meanings through his thick skull. Maybe you don't like your voice; maybe people have made fun of it. But it's the only voice you've got. It's your only source of power” (pp. 6-7). Elbow conveys that one’s idiosyncratic voice is ultimately what makes one’s writing interesting and effective, and I can certainly sympathize with what Elbow means here. In the past I have jokingly warned students that in looking over their writing, I am interested in their voices because AI-generated text is incredibly dull to read, and I would heavily imply a negative correlation between my level of boredom and student grades. However, while it’s true that many instructors find AI-generated text unstimulating, Tan and Matsuda remind us that an emphasis on a monophonic conception of voice privileges the values of the instructor in the learning process and does not prepare students for diverse, real-world rhetorical situations that may be satisfied by more formulaic, less inspired forms of writing (p. 231). If we know that such perspectives can potentially harm student learning, especially the learning of multilingual students, why are instructors now elevating voice as a standard for evaluating student writing? I would argue that fears surrounding AI and erasure of student voice are reflective of how the larger ideological function of writing coursework within the U.S. higher education landscape has evolved to confront the emergence of AI writing technologies. 

The beautiful desert landscape in Radium Springs, New Mexico.

The history of composition studies tells us that the first-year writing course was originally conceived as an institutional mechanism to sort and acclimate students into academic discourse. Robert Connors (1995) states that the first writing requirement was instituted at Harvard in the late 19th century because many incoming students were deemed unable to write proficiently, and first-year writing thus began as a remedial course intended to make up for the failures of secondary education. Arguing for the abolition of university writing requirements in the no longer active Journal of Advanced Composition, Sharon Crowley identifies the first-year writing course as executing composition’s ethic of service, where the course has historically served as “the site wherein those who are new to the academy learn to write its prose” (1995, p. 227). While Crowley acknowledges that there have been efforts to align the course with more democratic and liberatory aims, “a rhetoric of student need” articulated not by students or instructors but by more powerful academic stakeholders has thwarted these efforts, where fluency in SAE is conceived as necessary for student success in later university coursework and the professional world beyond. James Gee (1989) wrote that mastery of privileged dialects often relies on the acquisition of meticulous surface details, which take a long time to learn and thus efficiently indicate one’s membership in a particular discourse community. In her 1991 “Freshman English” essay, Crowley thus appropriately characterizes the project of first-year writing as having a gatekeeping function “deeply seated within cultural and academic expectations about who should have access to higher education” (p. 229).

The emergence of AI writing technologies challenges the original scope of composition’s regulatory function, as the affordances of AI make it easier for students to write fluently in SAE. If students are able to generate flawless SAE prose with a single ChatGPT prompt, how will institutions distinguish students who belong from those who don’t? Continuing within Crowley’s tradition, I would argue that voice replaces correctness as a new mark of otherness. While AI-generated prose is technically “correct” prose, much has been written about its derivative and generic style that is easy for instructors to detect. Sam Kriss (2025), in a timely New York Times Magazine article, characterizes AI-generated text according to its lexical (words like tapestry and groundswell), syntactical (“It’s not X, it’s Y” sentence structure), and mechanical (the em dash) features. Within the context of evaluating student work, this recognizable style conveys more than what meets the eye. While students are seemingly able to present the correct linguistic credentials, their voice is the tell of whether these credentials are forged. Instructor rhetoric surrounding the importance of developing one’s voice is merely code for discovering imposters within their midst, and as Tan and Matsuda (2026) highlight, quickly devolves into suspicion that students are having AI do the work for them. For instructors who consider any use of AI problematic, offending students will be punished for their deviance and the old system will have worked as intended.

Since instructors carry out the ethic of service for the institution and not students, Crowley questions whether the pernicious effects of university writing requirements can truly be reformed from the inside. In response to this question, Rebecca Moore Howard (1999) answers in the affirmative and offers guidance for our current predicament. While Crowley (1995) identifies composition’s ethic of service as ensuring that students write “error-free expository prose” (p. 227), Howard expands this concept beyond the mastery of language conventions and into the transgressive territory of plagiarism. As Crowley frames the role of composition as establishing a linguistic hierarchy of haves and have nots, Howard writes that acts of plagiarism, along with linguistic correctness, are an additional mark of otherness that college writing courses are required to root out. While most writing instructors would agree that penalizing student plagiarism is a worthwhile and necessary endeavor, Howard highlights that plagiarism constitutes a spectrum of behaviors with varying degrees of deviance and that students are being unfairly punished for the minor offense of patchwriting, in the same way they would be punished for “the purchase of term papers” (p. 87). A practice often employed by novice writers, patchwriting reflects a failed attempt to paraphrase, where writers make superficial changes to a source’s text rather than articulating the idea fully in their own words. Having dedicated much scholarship to this practice, Howard argues that acts of patchwriting stem from writers who do not fully understand what they have read and that they should not be treated as “ethical transgressors” but rather like students “laudably striving to learn” (p. 94). Suggesting that penalizing patchwriting is the easy way out for most instructors, Howard recommends that instructors should embrace the difficulty of grappling with and understanding the complexity of student learning processes and how these processes lead to behaviors like patchwriting. 

When it comes to the complicated pedagogical landscape of 2026, writing instructors should consider the historical implications of composition’s ethic of service when navigating concerns surrounding AI and student voice, and should strive to be conscientious about the potential impacts that our classroom policies and decision making are having on students. Who gets to decide what students need, and whose interests do our policies around AI serve? I am not necessarily against penalizing students for improper usage of AI, but want to encourage instructors to question whether our students share the same assumptions about writing that we do, and whether we should give students grace if our expectations are not initially aligned. In the same way that Howard is arguing for a middle ground, a healthy form of resistance requires us to be committed to the arduous task of distinguishing between appropriate and inappropriate behaviors that involve AI usage, rather than taking the easy path and banning all of its forms in our classrooms. While an AI prose style that surfaces in student writing may be representative of a careless student who mindlessly generated a writing assignment without much thought, it may also be indicative of a multilingual student who developed their ideas on their own and merely offloaded the proofreading process. For many of these scenarios, we can’t really know what happened until we engage in dialogue with students to not only understand how they are using the technology, but to also glean how their own sophisticated philosophies surrounding writing and authorship inform their writing goals. 

References

Connors, R. (1995). The new abolitionism: Toward a historical background. In J. Petraglia (Ed.), Reconceiving writing, rethinking writing instruction (pp. 3-26). Lawrence Erblaum Associates.

Crowley, S. (1995). Composition's ethic of service, the universal requirement, and the discourse of student need. Journal of Advanced Composition, 15(2), 227-239.

Crowley, S. (1998). A personal essay on freshman English. Composition in the university: Historical and polemical essays (pp. 228-248). University Of Pittsburgh Press.

Elbow, P. (1973) Writing without teachers. Oxford University Press. 

Howard, R. M. (1999). The New Abolitionism comes to plagiarism. In L. Buranen & A. M. Roy (Eds.), Perspectives on plagiarism and intellectual property in a postmodern world (pp. 87-98).

Kriss, S. (2025). Why does A.I. write like … that? The New York Times Magazine. https://www.nytimes.com/2025/12/03/magazine/chatbot-writing-style.html

Tan, X., & Matsuda, P. K. (2026). Using AI tools detracts from student voice ✨ Not every use of AI detracts from student voice. In C. Basgier, A. Mills, M. Olejnik, M. Rodak, & S. Sharma (Eds.), Bad ideas about AI and writing: Generative practices for teaching, learning, and communication (pp. 229-234). WAC Clearinghouse.