Generative AI is reshaping higher education, and our assessment practices must evolve to keep pace. Many institutions have introduced frameworks like traffic light systems, AI use scales, and mandatory declarations. These are helpful first steps because they give us language and structure while we find our bearings. But as Corbin, Dawson, and Liu (2025) argue in Talk is Cheap: Why Structural Assessment Changes Are Needed for a Time of GenAI, these approaches are limited; they rely on student compliance with unenforceable rules. The authors call these approaches discursive changes, or modifications that work through instructions without altering the tasks. Discursive changes alter the communication about the assignment, not the assessment itself. A simple example of discursive change is adding ‘GenAI use is not permitted in this assessment’ to existing assessment instructions.
The authors argue that discursive changes to assessments are well-intentioned but flawed because they assume students understand ambiguous rules and will comply even when non-compliance is advantageous. Discursive changes also work on the assumption that compliance can be verified, but current AI-detection tools are limited. The authors state, “current detection tools are fraught with false positives and negatives, creating uncertainty and mistrust rather than clarity and accountability” (p. 1092).
By contrast, Corbin, Dawson, and Liu argue that structural changes, “create assessment environments where the desired behavior emerges naturally from the assessment design” (p. 1093). In other words, structural changes modify the tasks, not the instructions. An example of a structural change provided by the authors include adding a “checkpoint in live assessment requiring tutor signoff on lab work.” Structural changes focus on the process, not the outcome. One example the authors offer: Rather than a final essay, students might participate in live discussions about their idea development and how their thinking developed based on feedback. Another structural change example includes designing assessments that connect throughout the term. Students build on their earlier work, demonstrating their learning across touchpoints, not from one task alone.
The authors conclude that long-term solutions require rethinking assessment design so that validity is built into the structure, not just explained in instructions. The challenge of assessment design continues as GenAI advances. Our time as educators is better spent on structural redesign of assessment to ensure assessment validity that demonstrates student capabilities.
Action item 1:
Have you modified your assessments with a structural approach? We’d love to hear from you! Join us for the AI Forum on Tuesday, Jan 27, 3:30 – 5:00 pm either in person or on Zoom to share your experience.
Action item 2:
After reading the article, consider the following questions:
-
- The article suggests shifting from product-focused to process-focused assessment. What “authenticated checkpoints” could you realistically build into your lessons or modules to capture a student’s developmental process?
- Think of an assessment you believe works well. What about the task itself encourages the kind of learning you want?
- What are barriers to making structural changes to assignments? How can we overcome them?
- Do you want to chat about this article? Send me an email (atallant@wcu.edu) and I’ll stop by your office or meet you on Zoom.
-
Action item 3:
Consider registering for CFC’s assignment re-design workshop in February!
Corbin, T., Dawson, P., & Liu, D. (2025). Talk is cheap: why structural assessment changes are needed for a time of GenAI. Assessment and Evaluation in Higher Education, 50(7), 1087–1097. https://doi.org/10.1080/02602938.2025.2503964
