These guidelines are not intended for specific implementation but rather to articulate principles for writing instruction in an academic environment where generative AI systems are widely available, unevenly understood, and increasingly normalized. They suggest pedagogical considerations for responding to AI without reducing writing to a set of transferable competencies, technical workflows, or compliance practices.

Rather than prescribing standardized uses of AI, these guidelines emphasize conditions under which writing remains a site of inquiry rather than production.

1. Writing as Risk, Not Mastery

As writers and faculty, we recognize that writing always exceeds our intention, control or optimization. As such, the goal of writing pedagogy cannot be the efficient production of stable, predictable texts (“output”).  Instead writing develops from our sustained engagement with uncertainty, difficulty, and non-coincidence between intention and outcome.

AI systems, which simulate coherence and ease, are not evidence of mastery but rather a demonstration of the emptiness of the mythology of mastery.

2. Non-Instrumentality of Writing Instruction

As always, writing courses are not sites of workforce preparation, tool proficiency, or institutional efficiency. They may be sites where the media and discursive operations of workplaces and institutions are investigated. However, we recognize that writing is not a neutral skill set that can be operationalized.

The presence of AI does not require faculty to reassert the utility of their work. Instead it invites reflection on what writing has never been reducible to. AI outputs are not writing; they are anticipations of what writing would appear to be. As such, the task of writing, whatever it may be, remains. 

3. Refusal of Theory as Application

Theoretical frameworks (e.g., posthumanism, rhetoric of technology, ethics of AI) should not be operationalized as explanatory tools or instructional scaffolds that resolve uncertainty. Theory presents interruptions rather than methods. Theory, broadly understood, cannot be applied to solve problems without gross distortion. This is the case with AI as well. Theory cannot insulate us from our exposure to AI. 

Approaches to AI in Writing Courses

4. AI as Provocation, Not Partner

AI systems should not be positioned as collaborators, co-authors, or assistants whose role is to improve student writing. Instead, they should be encountered as objects of inquiry that problematize writing itself.

Students may analyze, confront, or juxtapose AI-generated texts with their own writing, without an expectation of synthesis or resolution.

5. Authorship Without Recovery

Courses should not attempt to re-secure authorship through disclosure statements, process narratives, or documentation of human contribution. Such practices often reinscribe managerial control rather than ethical reflection.

Instead, writing instruction should acknowledge divided, distributed, and uncertain authorship as a condition rather than a problem to be solved.

Programmatic Considerations

6. Faculty Autonomy and Pedagogical Risk

Writing programs should protect instructor autonomy to experiment, refuse, or suspend the use of AI in their courses. No single AI policy or workflow should be mandated across sections.

Pedagogical inconsistency is not a failure but an expression of disciplinary seriousness. Variation and disagreement are signs of intellectual health rather than implementation problems.

7. Policy as Provisional

Any programmatic statements regarding AI should be explicitly provisional, subject to revision, and resistant to closure. Policies should articulate what is not yet known and what cannot be decided in advance. Include sunset clauses or built-in review periods for AI-related guidelines.