A recent UK survey reported that roughly 94 percent of undergraduates now use generative AI for assessed work, and the share of submissions containing AI-generated text rose from about 3 percent in 2024 to about 12 percent in 2026. Similar trends are visible on American campuses. When a majority of students use these tools, assessment designed for a different era may no longer measure what it claims to. Departments need a deliberate response rather than a patchwork of individual decisions.

Reacting with surveillance or bans alone has proven difficult. Detection tools have been pulled back at several universities because of false positives, and blanket bans are hard to enforce and easy to circumvent. Instructors who try to patrol every submission end up exhausted and adversarial. A more sustainable approach starts with asking what each assessment is for, and the departments that fare best usually begin by asking what a given assignment is meant to show, not by asking how to stop students from using tools.

A department-level review gives faculty a shared framework and avoids students facing wildly different rules in adjacent courses. It need not be elaborate. A few structured conversations and a simple audit of existing assignments can identify the highest-risk tasks and the best opportunities for redesign, and the audit can usually be completed in one or two meetings if faculty bring their syllabi and assignment lists, which leaves time for redesign work during the rest of the term.

Audit assignments by purpose and risk

List the major writing assignments in each course and note what learning outcome each is meant to assess. Mark which are most vulnerable to automation, such as generic essays on broad topics, and which already require engagement with course-specific material. For each vulnerable assignment, ask whether the outcome could be assessed in a more authentic way. The audit makes priorities visible.

• Audit each major assignment for its learning outcome and vulnerability to automation
• Add staged submissions, brief oral components, or in-class reflections
• State clearly where AI use is allowed and why
• Shift rubric weight toward reasoning and evidence
• Track results and share what works across the department

Redesign for visible thinking

Strengthen assessments by building in elements that show a student's own reasoning. Options include staged submissions with feedback, brief oral defenses, in-class reflections tied to the written work, and prompts that draw on class discussions, local data, or the student's own prior drafts. Each option makes it easier to see how the work developed. Choose the mix that fits your discipline and class size.

Be explicit about where AI use is acceptable and valuable. In some assignments, learning to use a tool critically, such as evaluating and correcting a generated summary, is itself the outcome. In others, independent writing is the target and tools should be restricted. Students respond better to rules that come with reasons, and writing the reason beside each rule on the assignment sheet removes much of the temptation to test the boundary.

Align rubrics and feedback

Review rubrics to make sure they reward the qualities that matter, such as reasoning, use of evidence, and engagement with sources, and not merely polished prose. Shift weight from surface features to substance where appropriate. Share the rubric with students and discuss it. Clear criteria reduce ambiguity and anxiety, and a quick comparison of old and new rubrics on a few sample papers shows whether the change in weighting produces the scores you intend.

Consider how feedback is delivered at scale. In large courses, where permitted, rubric-based tools can help produce consistent first-pass comments that instructors review and personalize, freeing time for conferences and oral components. Be transparent with students about this practice. Instructors remain responsible for every grade, and the same students who receive faster feedback often have more time to revise, which supports the learning outcomes the department cares about, while instructors keep final authority over every score.

Support faculty and monitor results

Provide faculty with resources, such as sample syllabus language, assignment templates, and a short guide to authorship conversations. Schedule time at department meetings to share what is working. Offer workshops during low-demand periods in the term. Change is easier when it does not add to an already crowded schedule, and a standing slot at one department meeting each semester for sharing assignment ideas keeps the work from being left to individual instructors.

Track outcomes. Compare student performance, completion rates, and instructor workload before and after redesign, and gather feedback from students and faculty. Share the findings and adjust. Treat the process as ongoing, since tools and student habits will continue to change, and a brief anonymous survey of instructors about workload and of students about clarity of expectations gives context that raw scores alone cannot supply to the department.