As generative AI reshapes writing, educators must rethink grading strategies. Explore four vital questions to enhance assessment integrity.
As we approach the fall semester, instructors find themselves facing a familiar yet daunting situation: receiving student submissions that may seem polished yet raise doubts about authorship. This moment brings to the forefront an essential reckoning for educators. The rise of generative AI tools has not necessarily introduced new problems; rather, it has unveiled vulnerabilities in our longstanding grading practices. Historically, the assumption that finished texts reliably reflect student effort and understanding has been a shaky foundation, one that has proven even more precarious with the advent of advanced AI.
It's time to rethink how we evaluate student writing. Rather than questioning the integrity of a submitted piece, we should ask what factors can ensure its trustworthiness from the outset. This shift leads us to consider four fundamental questions before assignments are given.
1. Is There Evidence of Student Thinking Beyond the Final Draft?
When only the final draft is submitted, it bears an impossible burden of proof. To address this, assignments should be structured to produce tangible evidence of student thought. Students could submit proposals, rough drafts showing their revision process, or reflection paragraphs detailing their changes. This method not only cultivates a sense of pride but also allows instructors to engage meaningfully with the student's growth. A transparent writing process naturally illuminates authorship, diminishing concerns about the source of the work.
2. How Does My Grading Rubric Align with Today's Writing Landscape?
Many existing rubrics were crafted in times when fluent and organized writing was a significant challenge, one that justified high marks. Today, fluency is increasingly accessible thanks to AI tools, making criteria based on mechanics and organization less relevant. To enhance assessment accuracy, it’s crucial to refocus rubrics on aspects that AI cannot easily replicate: a student's defense of their claims, the rationale behind their choices, or their ability to discuss feedback productively in conversation. If the criteria essentially reward what a machine can do, then the machine will undoubtedly excel in the grading process.
3. What Should My AI Policy Require from Students?
Simply asking students to declare whether they used AI offers little value in learning outcomes. Rather than a binary checklist, consider requiring students to document their decision-making processes regarding AI tools: what queries they submitted, the outcomes they received, and their reasons for keeping or rejecting specific elements. By framing the disclosure of AI usage in terms of critical decisions, students become engaged in their judgment and reasoning—skills we aspire to nurture in our teaching.
4. Who Should Handle the Difficult Cases?
Conflicts over authorship will arise; they always have. The key lies not in relying solely on AI detection scores or solitary instructors’ judgments late at night. The best resolutions come from collaborative discussions. Exchanging challenging submissions with a colleague for peer feedback can foster a more grounded deliberation. Spend half an hour discussing one ambiguous piece from each instructor. Each can independently provide a verdict with rationale, then compare insights. This exercise can unveil discrepancies in interpretation, highlighting areas where grading criteria might need refinement.
These four questions don't demand new technology or policies—what they require is a commitment to crafting assignments that yield valuable evidence, aligning evaluation criteria with evolving needs, and ensuring a human touch in the grading process. Even in large course formats, taking the time to offer genuine qualitative feedback remains incomparably more impactful than any algorithmic score. This human response forms the lasting memory of assessment for students.
As syllabi are finalized, consider addressing these pressing questions. Preparing these aspects early can lead to a more effective and authentic assessment process once students' work begins to arrive.
Roger Ochse, EdD, is a Professor and Honors Director Emeritus at Black Hills State University. With over forty years of experience in teaching college writing, he is developing the Green Pen Framework—a new approach to writing assessment focused on producing texts that are trustworthy through thoughtful assignment design.
Discussion
Sign in to join the discussion.