Process Over Product in College Writing

A Human-Centered Approach to Writing Instruction with AI

This morning, I found myself on the phone with my father, trying to articulate how several pieces of my sabbatical project were finally clicking together. I'd been mulling over Caryn Sever's insights from a recent "My EdTech Life" episode and Alan Liu's graduation address on writing in the age of AI. But it was the act of explaining these ideas to my father – that crucial human-to-human exchange – that helped crystallize my thinking. In fact, this conversation reinforced one of my core beliefs about teaching writing: while AI tools can enhance our pedagogy, meaningful human interaction remains absolutely essential to learning.

Two days ago, while hiking, I'd listened to an episode of "My EdTech Life" featuring Caryn Sever, where she described an innovative approach to using AI for student feedback. What struck me most was her insight about why this matters: employers are reporting that recent graduates often struggle to process critique, sometimes even "ghosting" their jobs after receiving critical feedback. Sever sees AI as one piece of a larger solution – a way to help students build the resilience and skills needed to engage with feedback productively. Her framework, though focused on coding assignments, creates an iterative feedback loop where students first get AI input (which feels less judgmental), then peer feedback (which builds community), and finally, instructor feedback. Each stage builds confidence and capability. This multi-step approach sparked an insight about the future of writing education that I've been circling around for months.

Here's what I'm beginning to see: The emergence of AI writing tools isn't just a challenge to overcome – it's an invitation to fundamentally rethink how we teach writing. For years, we've paid lip service to "process over product," but our assessment systems have remained stubbornly product-focused. Now, paradoxically, AI's ability to generate convincing essays might finally free us to fully embrace process-based assessment.

This isn't entirely new territory. When California community colleges still offered developmental education writing classes (before recent reforms eliminated them), we required comprehensive paper portfolios from our students. Students would turn in folders thick with evidence of their writing journey – free writes, brainstorming sheets, rough drafts, peer review notes – alongside their final drafts. We evaluated the whole package, not just the polished end product. But even then, the final draft carried outsized weight in our assessment.

In an age of sophisticated AI writing tools, final products are increasingly suspect. We could invest heavily in detection systems, but this approach is technically fraught and pedagogically destructive. Do we really want to create an atmosphere of suspicion between writing teachers and student writers?

Instead, what if we reimagined writing assignments entirely? What if, instead of providing a traditional prompt, instructors created a guided AI conversation system – essentially a specialized chatbot – that walks students through the writing process? The instructor would carefully engineer the prompts and scaffolding in advance, building in specific steps and decision points that guide students through topic exploration, research, drafting, and revision. Think of it as embedding our best writing pedagogy into an interactive tool that can give each student individualized attention while following a carefully structured path.

Picture this: A class exploring human creativity in the age of AI. The idea came from reading my former professor Alan Liu's graduation address on good writing in the ChatGPT era. Through his discussion of different types of creativity, I started thinking about how we might structure a writing assignment that encourages students to explore their own creative approaches. Students would begin by engaging with foundational readings about creativity as a class. I imagine lively discussions about what makes human creativity distinct – or perhaps not so distinct – from AI-generated work. These conversations would push students to wrestle with fundamental questions: What does it mean to be creative in an age of AI? How do human experiences, emotions, and contradictions shape our creative process in ways that might differ from algorithmic approaches? With this theoretical foundation in place, students would complete in-class, pen-and-paper exercises to develop their initial thoughts. Then, they would design a series of prompts that would help other students discover their own unique angle on the topic, whether that's through the lens of politics, cooking, automotive design, or poetry.

The instructor-designed chatbot would guide each student through a personalized exploration of the topic. Rather than leaving students to figure out the writing process on their own, the AI dialogue system would offer structured prompts that help them discover their unique angle – whether that's through the lens of politics, cooking, automotive design, or poetry. At each step, students would make choices that shape their path through the assignment, selecting their own examples and developing their own arguments while being supported by carefully crafted prompting sequences that embody best practices in writing instruction.

The chat might begin by asking the student to enter a meaningful quote and then guiding them through unpacking its significance: "Why did you choose this quote? How does it deepen our understanding of human creativity? What personal experience connects you to this idea?"

Students would test their ideas through multiple rounds of feedback, each serving a distinct purpose: AI feedback provides initial guidance in a low-stakes environment, peer review adds the crucial human element of audience awareness, and instructor feedback helps synthesize the insights gained. This layered approach, similar to what Sever advocates, helps students develop both their writing skills and their ability to engage productively with critique – a vital skill for their academic and professional futures. The final submission would include their handwritten worksheets from class activities, a record of their AI conversation (via a shareable URL), their peer review notes, and yes, a final essay – but that essay would be almost an afterthought, a natural culmination of the process rather than its primary focus.

This approach to integrating AI instruction into the writing process will become easier to implement as higher education institutions offer free access to quality AI tools as the California State University system has just done. More crucially, this approach aligns with our understanding of how individuals learn to write: primarily through exploration, dialogue, and iterative processes.

Let me be clear: I'm not advocating to replace human interaction with AI. Just as my own thinking crystallized through conversations with my father, students need those vital exchanges with peers and instructors. In-class discussions, peer review sessions, and face-to-face conversations with instructors are irreplaceable moments where ideas deepen and understanding grows. What I'm suggesting is a balanced approach in which AI becomes one tool among many, supporting—but never replacing—the human elements that make writing education transformative.

The future of writing education might not be about fighting AI or pretending it doesn't exist. Instead, it might be about embracing AI as a tool for deepening the very human process of developing and articulating ideas. In doing so, we might finally build the truly process-focused writing education we've been talking about for decades.

---

This post was created in collaboration with AI tools, reflecting my commitment to understanding both the potential and limitations of AI in writing instruction.