Rethinking Assessment in the Age of Generative AI
Authors: Yerin Seung; info@ciddl.org
As generative AI (GenAI) tools become increasingly integrated into everyday learning environments, educators are facing a dilemma: How can we ensure that our assessments continue to reflect students’ authentic understanding and skills? Many institutions have responded by crafting detailed rules and guidelines that specify when, how, and under what circumstances students may use AI. These often include policy statements, AI use declarations, or rubrics clarifying permissible use. While helpful for promoting awareness, these approaches rely entirely on student compliance, an increasingly unrealistic assumption.
According to Corbin and colleagues (2025), what we need is not just more explicit instructions about AI use, but a fundamental shift in how we design assessments themselves. In other words, talk is cheap; it’s time to reevaluate the structure of assessment in an AI-empowered education.
The Limits of Rule-Based Approaches
Discursive changes to assessment refer to modifications of assessment that rely solely on the communication of instructions, rules, or guidelines to students, rather than relying on other methods. Statements like “AI may be used for brainstorming but not drafting” or requiring students to disclose their AI use in a written declaration fall under discursive changes. Corbin and colleagues (2025) identified that these strategies are based on three shaky assumptions:
- Students fully understand what is allowed.
- Students will voluntarily follow the rules.
- Educators can verify whether the rules were followed.
In reality, AI use is hard to define, easy to conceal, and nearly impossible to detect. The more detailed our rules become, the more obvious it is that we can’t enforce them. This creates a paradox: students with good faith are burdened by compliance, while others bypass the system undetected. Relying solely on discursive changes creates a fragile assessment ecosystem.
Why Structural Change Matters
In contrast, structural changes to assessment do not entirely depend on telling students what not to do. They reshape the task so that the design itself either limits inappropriate AI use or transforms it into a meaningful part of the learning process. These changes share a key principle: they modify the assessment task itself rather than relying on instructions surrounding it. Here are two powerful structural strategies educators can begin implementing:
1. Shift from Product-Focused to Process-Focused Assessment
One major risk of GenAI is that students can produce high-quality final products with minimal engagement. Structural redesigns address this by making the process, not just the outcome, visible and assessable.
This might involve:
- Requiring students to submit drafts over time, annotated with reflection on how their thinking evolved
- Including peer feedback checkpoints that track revision and growth
- Holding live or recorded discussions where students articulate their reasoning or critique their own earlier work
These process-focused assessments offer a more reliable window into students’ actual learning. They also shift the emphasis from what students submit to how they develop and apply their knowledge.
2. Build Validity Across Interconnected Tasks
Trying to make each assignment “AI-proof” is becoming increasingly futile. Instead, educators can take a broader view—designing interconnected assessments across a unit or module that build on each other.
For example:
- A project proposal submitted early in the term could later be referenced or revised in a final report
- A presentation might be based on data or analysis from earlier tasks, making it harder to disguise disconnected AI use
- An in-class reflection could require students to explain decisions made during a take-home assignment
This approach treats assessment as an evidentiary chain, where validity stems not from a single isolated task, but from a coherent demonstration of learning across multiple, strategically linked activities (St-Onge et al., 2017).
What Educators Can Do Now
Rethinking assessment structure does not mean discarding everything we’ve done before. It means intentionally revisiting what each task is designed to measure and how students demonstrate that learning.
Here are a few starting points:
- Identify assessments in your course that could be completed entirely with GenAI, and consider redesigning them to include in-person components, process tracking, or multiple checkpoints
- Use AI as part of the task, but assess how students interact with, critique, or improve upon its outputs
- Develop course-level progression where earlier work must be referenced or justified in later submissions
There is no one-size-fits-all solution for this. Structural change takes time, planning, and alignment with learning outcomes. But they offer a more sustainable path forward than relying on unenforceable policies or declarations.
From Compliance to Design
The rise of GenAI challenges us not just to police behavior, but to rethink what learning and assessment should look like. Instead of trying to control AI use through instructions alone, we can design assessments that embrace authenticity, process, and coherence. If we want our assessments to reflect real learning in an AI-enhanced world, we need more than words—we need structure.
Join the conversation in our community!
CIDDL is committed to providing high-quality resources to support the increasing knowledge, adoption, and use of a range of educational technologies that can be used for educators, related services, or leadership preparation programs. For more resources, including videos and blogs, subscribe to our newsletter and follow us on YouTube, Facebook, and LinkedIn. The most important part of our CIDDL community is YOU. Join our community and share the innovative ways you are using technology, ask a question about technology integration, or participate in our bi-weekly live AI Community Chats. We look forward to seeing you in our community!