AI-Resilient Assessment Redesign

Lorri Barnett

Show / hide video transcript
– Hi, I’m Lorri Barnett. I am the Senior Director of Professional Development for Quality Matters, which is based in the United States. Quality Matters is a research-based framework for designing high quality online and hybrid courses. And the activities in this session use QM principles to strengthen alignment, clarity, and student success. I am also an adjunct faculty for Purdue Polytechnic in West Lafayette, Indiana, which is considered the Midwestern part of the United States. In this activity, we will demonstrate a pattern that I started seeing in my online courses: polished, flawless assignments that didn’t sound like the same student that I was seeing in other written assignments or discussions.
This activity will help educators move from trying to catch AI to designing assessments that will still provide valid evidence of student learning in an AI-rich environment, aligning with QM’s emphasis on valid, reliable assessments that measure stated objectives, which happened to be our General Standard 3, and also upholds academic integrity. Initially, when I began looking at this worksheet, I was focused on just catching students using AI, but that really wasn’t sustainable or aligned with the culture and environment that I was trying to instill in the classroom. I wanted to turn to my design standards. What is this assignment actually measuring, and how can I see students thinking?
So we will be using a QM-informed worksheet to clarify outcomes, to design the process, and to make criteria around AI expectations more transparent. The goal of this activity is to make AI a tool within a clear structure, not the shortcut that replaces learning. The purpose of our AI-Resilient Assessment Redesign Worksheet is to help faculty redesign one assignment so that it emphasizes authentic learning, process, and transparent criteria in an AI-rich environment. The concept proposal identifies three design moves: make outcomes explicit, design for the process, and use transparent criteria. This worksheet is grounded in the QM Higher Education Rubric, and it helps faculty apply QM principles to assessment design in an AI-rich environment. It especially draws on our General Standard 2, which are our ‘Learning objectives and competencies’; our General Standard 3, which is “Assessment and Measurement”; our General Standard 5, which is “Learning Activities and Learner Interaction”; General Standard 6, which is “Course Technology”; and there is some relevance to General Standard 4 around “Instructional Materials” as well. This worksheet is going to be around the example I chose from my own work: “Introduction to Environmental Policy.” That’s from the demonstration that I want to show you. The original assignment was, “Students were to write a 1000 word paper on current environmental issues.” So this worksheet has you look at “before,” connecting the QM standards, and any revision that might be needed. And this is around trying to incorporate good use of AI inside your students’ outcomes and submissions.
So the learning outcome before was simply “Understand environmental policy issues.” Pretty vague, pretty broad, right? This aligns to our QM Standard 2, especially 2.1, 2.2, and 2.3, on clearly stated measurable objectives. So notice that is a pretty broad objective stated there. So the revision that I needed to make was, “Compare two policy responses to a current environmental issue and justify which is more effective using evidence from credible sources.” The credible sources is what opens the opportunity for students to utilize the use of AI inside of their work.
Then we look at the assessment task. I had written that “Students must submit one final paper at the end of the week.” The prompt is too broad and can easily be completed with generic AI output because I’m not specific in what I asked them to do. This aligns to our General Standard 3.1 on assessments, and then measuring against 3.4. And then, finally, sequencing those assessments, 5.1. So what I decided to do was to break the assignment into stages where there would be a topic proposal that they would submit, a source comparison chart, and a draft paragraph with their reasoning, and final submission. By allowing students to provide drafts in chunks, it allowed me to incorporate the use of AI with them, mentor them, and give them good, solid use of AI sightings and sources inside of their work.
Process evidence is the next category, and it sort of lends itself very well to the prior one. At first, I did not have a draft. There were no planning notes. I did not ask for exclamations from students on how they developed their work. It was one final submission that they were to complete. Alignment to QM is against General Standard 3.5 and 5.2. And the way that I made my revision was I required a short reflection explaining how their sources were selected, what revisions did they make, and whether AI tools were used for brainstorming, outlining, or editing. And by simply allowing their creativity to include and foster the use of AI, it changed the dynamic of the assignment and the outcomes 100%. And that was in a very positive direction as well.
My evaluation criteria was the next segment of my paper that I looked at my outline. My grading was based on that rubric mostly on grammar, the length of the paper, on formatting, with limited attention to evidence or reasoning, which is key in this assignment. This aligned to our General Standard QM 3.3 on descriptive and evaluation criteria. So what I did was change my rubric. I added a rubric to emphasize quality, comparison, policy options, use of evidence, justification of choice, and reflection on their decision making. What allowed me to really beef up the rubric was also giving them feedback along the way on the segments that I had them submit for draft review. And they would utilize the comments that I would give them based on, “You’re using too much AI. You did not cite. You did not source. I need you to explain to me your justification and reasoning for why you selected that piece of evidence.” And that helped them come up with a final product that they could be proud of and that incorporated AI in the right way, the use of AI in the right way.
Academic integrity guidance. The assignment directions did not explain what counted as acceptable or unacceptable use. I talk about AI and the use of AI in the syllabus in a very broad spectrum. I talk about use, how they may use it, how they must cite it, but I did not dive into the details that I feel are necessary inside each assignment. This aligns to QM’s Review Specific Standard 3.6. And how I revised my assignment was I added guidance to the assignment itself. I specified what they might be allowed to use, how to acknowledge that use of AI, and what parts of the analysis must remain in their own words. And that helped seal the deal, if you will, on doing a very effective use of AI, of research in a new way, and giving the flexibility to students that they were so looking for in terms of doing great work.
Finally, on the course technology piece, I had, “They may use these digital tools,” but I did not give them guidance on the learning attachment to the goals and the outcomes. So it was one thing to be very disconnected in telling them they could use these tools, but quite another to align that to an actual learning outcome. That is aligned to our 6.1 and our 6.2 General Standard. And I was specific in naming which tools they could use. So I like to use NotebookLM, Claude, and ChatGPT. And so, I absolutely was explicit in outlining the tools I wanted them to use, how to use them, and how to use them to strengthen and not replace their own words and their own thoughts. And again, this redesign helps you apply our rubric to one of today’s most urgent teaching questions: how do we create assessments that are still measuring authentic learning when students have such access to AI tools all around them? This worksheet helps you keep that focus on measurable objectives, aligned assessments, transparent criteria, active learning, and clear academic integrity guidance. The specific standards that we used are listed there. And if you want more information on Quality Matters Higher Education Rubric, please reach out to [email protected], and someone will be happy to assist you. Thanks for working through this activity with me.
If you found it helpful, I encourage you to complete the worksheet for one course that you’re teaching now and just notice what changes for your students. OneHE has more microlearning resources, and Quality Matters offers deeper frameworks and peer review support to help strengthen your ongoing course design work. The worksheet resources are contained below this video.
ACTIVITY PURPOSE
Assignments that AI tools can complete convincingly are one of the most common sources of assessment anxiety in higher education. In this activity, you will redesign one assignment to shift the focus from product to process so that AI alone can’t do the work, and, instead, students must interpret, justify, reflect, or apply the course materials in context. Students learn how to use AI, if allowed, in ways that support rather than replace their learning. By clarifying outcomes, emphasizing process, and making criteria transparent, you design tasks where AI can be a support tool but cannot replace students’ own reasoning and decision-making. When completed, you will have a redesigned version of one assignment, plus a short paragraph of guidance you can share with students explaining what you’re looking for and where AI fits in.
This activity was designed with Quality Matters (QM). QM is a non-profit, quality assurance organization whose Higher Ed Rubric is one of the mostly widely used to underpin the quality of course delivered online synchronously and asynchronously and for hybrid, Hyflex, and multi-modal courses. This activity applies one of its core principles — making student learning visible — to a challenge every educator is facing today: designing assessments that hold up in an AI-rich environment. View Specific Review Standards from the QM Higher Education Rubric (7th Edition). Access to the full, annotated rubrics is available to QM members only. For more information about becoming a QM member, please visit Quality Matter Membership or email [email protected]. Learn more about OneHE and QM partnership.
USEFUL FOR
This activity is particularly well suited to educators who are:
- Concerned that AI tools could complete one or more of their assignments without genuine student learning taking place
- Looking for a constructive, design-based response to AI rather than a policing or detection-based one
- Reviewing or refreshing assessments as part of a course redesign or quality review process
- Wanting a practical starting point before writing or updating institutional AI guidance for students
- Working in departments or teaching teams looking for a shared approach to assessment redesign
PREPARATION
Have the current instructions (and rubric, if you have one) for the assignment you want to redesign to hand, along with a blank document where you can build your table or use the downloadable worksheet. No other preparation is needed.
INSTRUCTIONS
Before you begin: Choose one assignment where you are most concerned about undetectable AI use, for example, a standard essay, short-answer exam, or take-home report.
Download an AI-Resilient Assessment Redesign Worksheet (Word, 336 KB) or have a blank document open alongside the current version of the assignment instructions.
- Step 1. Set up your before/after table. Using the worksheet or in your blank document, create a table with two columns, “Current Version” and “Redesigned Version”, and six rows:
- Row 1: Learning outcome (what you want students to learn)
- Row 2: Assessment task/instructions (what students are told to do)
- Row 3: Process evidence (what you look for when grading)
- Row 4: Evaluation criteria (how the work will be judged/scored)
- Row 5: Academic integrity guidance (how AI could complete this task without real learning)
- Row 6: Course technology (what platforms or tools are used to deliver, support, or monitor this assignment)
- Step 2. Fill in the “Current Version” column
- Learning outcome: in your own words, state what this assignment is supposed to teach or measure
- Assessment task/instructions: paste or summarise what students are currently asked to do
- Process evidence/evidence of learning: list what you truly value when grading, for example, reasoning, use of sources, or application to context
- Evaluation criteria: outline how the work will be judged or scored, for example rubric categories, weighting, or the standards that separate a strong response from a weak one
- Academic integrity guidance: describe how AI could complete this task without genuine learning, and where the gaps or risks are that your current design doesn’t catch
- Course technology: list the platforms, tools, or software used to deliver or support this assignment, for example LMS, plagiarism/AI detection tools, or specific apps students must use
- Step 3. Redesign using three design moves
- Design move 1: Make outcomes explicit and granular.
Rewrite the assignment purpose with one or more specific outcomes using observable verbs, for example, “compare three scholarly sources and justify which argument is strongest based on evidence quality”. Make sure students can see these outcomes. - Design move 2: Design for process, not just product.
Revise the task instructions to require students to show their process. Options include: requiring annotated drafts or planning notes; asking for a short process reflection: “What steps did you take? Where did you use AI, if at all?”; building in checkpoints, such as a proposal, outline, or peer feedback, that lead to the final product. - Design move 3: Use transparent criteria.
Create or revise a rubric that rewards the thinking you want to see, such as quality of evidence, justification of choices, or integration of feedback, rather than only surface features like length and grammar. Share this rubric with students before they begin.
- Design move 1: Make outcomes explicit and granular.
- Step 4. Write a short “Student Guidance” paragraph.
Explain to students:- What the assignment is helping them learn
- How they may or may not use AI tools
- What evidence of their own thinking you’ll be looking for
Once you implement the redesigned assignment, use the new rubric to look specifically for evidence of process and critical thinking. Consider adding a brief reflective component where students describe how they approached the task, including if and how they used AI. You can grade this reflection lightly but use it to gauge whether students are engaging with the assignment as intended.
DURATION
20–30 minutes to complete the worksheet and draft your redesigned assignment and student guidance paragraph. Building or revising a full rubric to match your redesign may take a little longer, depending on how detailed you’d like it to be.
ADAPTATIONS AND EXAMPLES
This worksheet works well for written assignments such as essays, reports and exams, but the same three design moves apply to other assessment types too. For lab reports or problem sets, focus the “process” move on requiring worked steps, calculations, or annotated data rather than just a final answer. For presentations or creative work, the process move might mean asking for planning notes, drafts, or a short reflection on choices made.
If you’re redesigning a group assignment, add an extra row to your table for “individual accountability”, and use the guidance paragraph to clarify how individual contributions will be identified and assessed.
This activity also works as a shared exercise in a department meeting or teaching and learning workshop: colleagues can redesign different assignments in parallel, then compare their “before” and “after” versions and share design moves that worked well.
Evaluating your redesign
You will know this redesign is working when:
- It becomes harder for AI-generated work to meet your rubric without genuine understanding
- Student work shows more individual voice, reasoning, and connection to course concepts
- Students can explain what they learned from the assignment beyond “finishing the paper”
- Your concerns shift from “Did AI write this?” to “What does this show about the student’s thinking?”
Compare samples of student work from before and after the redesign to see whether the new version gives you clearer evidence of learning.
TECHNICAL REQUIREMENTS
A word processor or document editor capable of creating a simple table (e.g. Word, Google Docs). No other tools or software are required.
RELATED RESOURCES
- Download an AI-Resilient Assessment Redesign Worksheet (Word, 336 KB)
- Quality Matters. (2023). Quality Matters Higher Education Rubric. (7th ed.). — The rubric that underpins the alignment, clarity and transparency principles used in this activity.
- Ambrose, S. A., et al. (2010). How Learning Works: Seven Research-Based Principles for Smart Teaching. Jossey-Bass. — A foundational text on how students learn; useful background for understanding why process-focused design supports genuine learning.
- Your institution’s teaching centre or CTL guidance on AI and assessment design — most institutions now publish their own guides, which are worth checking alongside general principles here.
Explore similar practical activities from Quality Matters:
Have feedback?
We’d love to hear what you think about this activity. If you have any feedback or would like to share how it has helped you, please email us at [email protected].
