Evidence & Assessment Design
Design your assessment strategy, build performance-based and supporting assessments, create rubrics, and verify alignment with your learning outcomes.
Learning Outcomes
By the end of this module, you will be able to:
- ✓ Distinguish between performance-based/authentic assessments and traditional/supporting assessments
- ✓ Design authentic performance tasks using the GRASPS framework and other approaches
- ✓ Create rubrics that clearly communicate expectations
- ✓ Use AI to generate and refine various assessment types
- ✓ Verify alignment between your learning outcomes and assessments
- ✓ Design AI-resistant and AI-integrated assessments appropriate for your context
Design Your Assessment Strategy
Stage 2 asks: “How will we know if students have achieved the desired results?”
Before planning lessons, we design assessments that will reveal whether students truly understand—not just remember facts, but can apply, analyze, and transfer their learning. In UbD, assessments fall into two complementary categories:
Inclusive Assessment Is Good Assessment
Before designing specific assessments, consider this: assessment barriers often measure the wrong things. If a student understands the material but can’t demonstrate it through a single, rigid format, the assessment is measuring the format—not the learning.
Universal Design for Learning (UDL) ↗ offers three principles that strengthen any assessment design: CAST. (2018). Universal design for learning guidelines version 2.2. CAST.
• Multiple means of engagement — Offer choice in how students engage with the assessment (e.g., topic selection, collaboration options, real-world connections) so motivation and investment are supported.
• Multiple means of representation — Present assessment instructions, criteria, and models in more than one format (e.g., written rubrics plus verbal walkthrough, sample exemplars) so all students understand what’s expected.
• Multiple means of action and expression — Allow students to demonstrate their learning in different ways (e.g., written report, oral presentation, video, or portfolio) so the assessment captures understanding, not just one mode of communication.
You don’t need to redesign everything. Even small adjustments—like offering a choice of product format or providing rubrics in advance—make assessments more inclusive. Throughout this module, look for opportunities to build in flexibility.
Two Categories of Assessment
Authentic Assessment
Complex, authentic tasks where students apply learning to realistic situations. Provides the strongest evidence of deep understanding and transfer.
Traditional Assessment
Assessments that check foundational knowledge and skills along the way, supporting learning toward the larger performance tasks.
A Note on Terminology: Authentic Assessment vs. Performance Tasks
In the broader educational development community, the widely recognized term for complex, real-world assessments is authentic assessment—any assessment that asks students to apply their learning to realistic, meaningful tasks.
Performance tasks are a specific type of authentic assessment used in the Understanding by Design (UbD) framework, often structured using the GRASPS model. But authentic assessment is a much larger category that also includes case studies, portfolios, simulations, oral exams, and many other approaches.
Throughout AI DesignLab, we use "authentic assessment" as the broad category and "performance tasks" when referring specifically to GRASPS-style structured tasks. You may encounter either term in educational literature—they are closely related, and the design principles are the same.
Both Play Different Roles
Authentic assessments provide the strongest evidence of deep understanding and transfer. Traditional/supporting assessments check foundational knowledge and skills along the way, when needed. Most courses use a combination of both, but performance-based tasks should be the primary evidence of achievement.
Assessment Strategy Map
Here’s how the pieces of your assessment strategy connect. We’ll work through each step in the sections that follow.
Assessment Strategy
Decide which outcomes need performance-based vs. supporting assessment.
Build Assessments
Design performance tasks and supporting assessments for your outcomes.
Create Rubrics
Develop clear criteria and descriptors for evaluating student work.
Think Like an Assessor
What would convince you that students really “get it”? What evidence would show they can use their learning in meaningful ways? Start with that question, then choose the assessment type that best captures that evidence.
Authentic Assessment
Authentic assessments ask students to demonstrate understanding through authentic, complex tasks. These provide the strongest evidence that students can transfer their learning to real-world situations. They include both structured frameworks like GRASPS and a wide range of other authentic approaches.
The GRASPS Framework
The GRASPS ↗ framework helps you design authentic performance tasks that put students in realistic roles solving meaningful problems.
Example: Biology Course
🎯 Goal
Explain how a disease disrupts cellular function and recommend a treatment approach.
👤 Role
You are a medical researcher presenting to a hospital board.
👥 Audience
Hospital administrators who need to understand the science but aren't experts.
📍 Situation
The hospital is deciding whether to fund a new treatment program.
📦 Product
A 10-minute presentation with visual aids and a 2-page executive summary.
📏 Standards
Scientific accuracy, clear explanation for non-experts, persuasive evidence.
Beyond GRASPS: Other Authentic Assessment Approaches
GRASPS is one powerful framework, but authentic assessment takes many forms. Consider these approaches, which can be used alongside or instead of GRASPS tasks. They accommodate diverse forms of expression, reduce test anxiety, and provide authentic demonstrations of competence.
AI Prompts for Authentic Assessment Design
Design an alternative assessment for my course. Course: [Course name] Learning Outcomes: [What should students demonstrate?] Assessment Type: [Oral exam/Video/Concept map/Peer teaching/Journal/Other] Class Size: [Number of students] Constraints: [Time, technology, accessibility considerations] Please provide: 1. Detailed Assignment Description - What students will create/do - Format specifications - Length/duration requirements 2. Student Instructions - Step-by-step process - Resources they can use - Timeline with milestones 3. Assessment Rubric - 4-5 criteria with clear descriptors - Both content and format/delivery considerations 4. Logistics Plan - How you'll manage the assessment - Scheduling considerations - Technology requirements 5. Accessibility Accommodations - Alternative options for diverse learners
Generate oral examination questions for a viva/oral defense. Subject: [Topic area being examined] Context: [Project defense/Comprehensive exam/Skill demonstration] Duration: [How long is the oral exam?] Level: [Undergraduate/Graduate/Professional] Learning Outcomes: [Paste the learning outcomes this oral exam should assess] Key Topics / Content Areas: [List the major topics or content areas to be covered, e.g., "Cell biology, genetics, evolution" or "Patient assessment, pharmacology, clinical reasoning"] Please ensure questions are aligned with the learning outcomes and cover the key topics listed above. Please create: 1. Opening Questions (warm-up, lower stakes) - 3-4 questions to ease into the exam 2. Core Knowledge Questions - 5-6 questions testing fundamental understanding 3. Application Questions - 4-5 questions requiring students to apply knowledge to scenarios 4. Critical Analysis Questions - 3-4 questions probing deeper thinking and synthesis 5. Follow-Up Probes - Questions to dig deeper based on responses - "Can you elaborate on...?" - "What if the situation were different...?" 6. Closing Questions - Reflection and self-assessment opportunities Include suggested time allocation for each section.
Traditional / Supporting Assessment
While authentic assessments provide the strongest evidence of understanding, traditional/supporting assessments play an important role in checking foundational knowledge and skills along the way. They help you and your students track progress toward the larger performance tasks.
Types of Supporting Assessments
Quizzes & Tests
Check factual knowledge, terminology, and foundational concepts. Most effective when used as low-stakes formative checks rather than high-stakes gates.
Discussions & Participation
Assess students’ ability to articulate ideas, engage with peers, and think on their feet. Can be structured or informal.
Formative Checks
Quick, low-stakes activities that reveal student understanding in real time: exit tickets, polls, one-minute papers, muddiest point.
Homework & Practice
Assignments that build skills and reinforce learning between classes. Most effective when they scaffold toward performance tasks.
Supporting Assessments Should Scaffold
The best traditional assessments don’t just check knowledge—they build toward your performance tasks. Ask yourself: “Does this quiz/homework help students prepare for the bigger assessment?” If not, reconsider whether it’s needed.
Building Effective Rubrics
Rubrics make expectations clear and grading consistent. A good rubric describes what quality looks like at each level for each criterion—using language that is specific, distinct, and observable.
How to Build a Rubric: Start from the Top
Building a rubric is easier when you follow a structured, top-down approach. Rather than trying to fill in every cell at once, work through these steps:
Step 1: Identify Your Criteria
Start with your learning outcomes and performance task. Ask: What are the 3–5 most important dimensions of quality in this work? Each criterion should represent a distinct, assessable aspect of student performance (e.g., "Use of Evidence," "Technical Accuracy," "Communication Clarity").
Step 2: Define the Top Level First
For each criterion, describe what excellent performance looks like in concrete, observable terms. Be specific: instead of "sophisticated analysis," describe what that analysis actually includes (e.g., "Identifies at least three causal relationships and explains how they interact"). This top-level description becomes the anchor for all other levels.
Step 3: Work Downward
Once you have the top level, define each lower level by asking: What is missing, reduced, or less developed compared to the level above? Each level should be clearly distinguishable—a reader should be able to tell which level a piece of work falls into without guessing. Avoid vague qualifiers like "adequate," "appropriate," or "good"; instead, describe what you would actually see in the student's work.
Sample Rubric: Specific, Observable Descriptors
Notice how each cell in this sample describes what you would actually see in the student's work. The levels are distinct—you could place a student's work at a specific level without debating which one it fits.
| Criteria | Exemplary(4 pts) | Proficient(3 pts) | Developing(2 pts) | Beginning(1 pt) |
|---|---|---|---|---|
| Scientific Accuracy | All key concepts are factually correct; identifies at least three relationships between concepts and explains how they interact | All key concepts are factually correct; identifies relationships between concepts but does not explain interactions | Most key concepts are correct (1–2 minor errors); identifies some relationships but connections are incomplete | Contains three or more factual errors; concepts are presented in isolation without identifying relationships |
| Communication | All technical terms are defined on first use; explanations use analogies or examples that connect to the audience's experience | Most technical terms are defined; explanations are clear but do not include analogies or examples for the audience | Some technical terms are left undefined; explanations are present but assume prior knowledge the audience may not have | Technical terms are used without definition; no attempt to adapt language or explanations for the intended audience |
| Evidence & Reasoning | Every claim is supported by at least one cited source; reasoning explicitly connects evidence to the claim it supports | Most claims are supported by cited sources; reasoning connects evidence to claims but connections are implicit in places | Some claims are supported by evidence; at least two claims lack cited sources or the reasoning jumps from evidence to conclusion | Claims are stated without supporting evidence; no sources are cited, or sources are present but not connected to specific claims |
Watch for Vague Descriptors
Terms like "sophisticated," "appropriate," "adequate," and "good" feel natural but don't tell students (or you) what to look for. Compare: "Appropriate use of evidence" vs. "Every claim is supported by at least one cited source." The second version gives students a clear target and makes grading consistent across evaluators.
Rubric Best Practices
Build top-down: define excellent first, then work down by identifying what's missing at each level. Focus on observable features of student work, not inferred qualities. Share rubrics with students before the task—they're learning tools, not just grading tools!
Verify Outcome–Assessment Alignment
Before moving on, take a moment to verify that your assessments actually measure your learning outcomes. Misalignment is one of the most common course design problems—and it’s easy to fix now, before you plan learning activities.
Alignment Check Questions
- ✓ Does every learning outcome have at least one assessment that measures it?
- ✓ Are your most important outcomes assessed through performance-based tasks (not just quizzes)?
- ✓ Do your supporting assessments scaffold toward the performance tasks?
- ✓ Are there any assessments without a clear outcome they measure? (If so, reconsider them.)
Use AI to Check Alignment
In the AI Activities section below, you’ll find a prompt specifically designed to help you verify alignment. Paste your learning outcomes and assessment descriptions, and AI can identify gaps or redundancies. But remember—your professional judgment is the final check.
Supporting Student Success
Well-designed assessments only work if students understand why they're being assessed that way and have opportunities to practice before the stakes are high. This section covers two strategies that support student buy-in and reduce assessment anxiety: assessment justification and low-stakes practice design.
Assessment Justification: Explaining the "Why"
Students are more engaged with assessments when they understand the reasoning behind them. Rather than simply assigning a task, take a moment to explain why this particular assessment format was chosen and how it connects to their learning and future goals.
Connect to Outcomes
Show students exactly which learning outcomes the assessment measures and why those outcomes matter. "This case study assesses your ability to [outcome]—a skill you'll use every day in clinical practice."
Explain Real-World Relevance
Connect the assessment format to authentic professional tasks. "Practitioners in this field regularly write policy briefs like this one for non-specialist audiences."
Address the Format Choice
If you're using a non-traditional format, explain why. "I chose an oral defense instead of a written exam because the ability to explain your reasoning under questioning is critical in this profession."
When to Justify
Include a brief rationale in the assignment description itself, discuss it when introducing the assessment in class, and revisit it when returning feedback. Students who understand the "why" are more likely to engage deeply rather than just chase points.
Low-Stakes Practice Design
Students perform better on high-stakes assessments when they've had opportunities to practice the same skills at lower stakes first. Low-stakes practice reduces anxiety, builds confidence, and gives both you and your students early feedback on their progress.
Completion-Based Grading
Grade early practice attempts on completion and good-faith effort rather than quality. This encourages students to take risks and learn from mistakes without fear of grade penalties. Reserve quality-based grading for later submissions when students have had time to develop skills.
Revise-and-Resubmit
Allow students to revise and resubmit work after receiving feedback. This shifts the message from "you failed" to "you're not there yet—here's how to improve." Set clear boundaries: one revision cycle, within a defined timeframe, with specific feedback to address.
Scaffolded Drafts
Break major assessments into smaller checkpoints with feedback at each stage: an outline, a draft section, peer review, then the final submission. Each checkpoint is low-stakes; the final product is the high-stakes assessment.
Ungraded Practice Tasks
Offer practice versions of high-stakes assessments—sample case studies, practice quiz questions, or mock presentations—that mirror the real task without carrying grade weight. Students can self-assess or receive peer feedback.
Connecting to Your Assessment Plan
Look at the supporting assessments you designed (or will design) in the AI Activities section. Which of them could serve as low-stakes practice for your performance task? Could you grade early versions on completion and later versions on quality? Could you build in a revision opportunity for your major assessment? These aren't extra work—they're strategic uses of assessments you're already planning.
Optional Prompt Add-Ons
Use these prompts to enhance any assessment you've designed. They're optional additions you can incorporate into your assessment descriptions or supporting materials.
Help me write a student-facing rationale for this assessment. Assessment description: [Paste your assessment description] Learning outcomes it measures: [Paste the relevant outcomes] Why this format: [Brief note on why you chose this format over alternatives] Please write a 2-3 paragraph explanation I can include in the assignment description that: 1. Explains which learning outcomes this assessment measures and why they matter 2. Connects the assessment format to real-world professional practice 3. Describes how this assessment prepares students for their future careers or further study 4. Uses encouraging, student-friendly language (not defensive or apologetic) Keep the tone warm and motivating—help students see this as an opportunity, not just an obligation.
Help me design low-stakes practice activities that prepare students for this assessment. High-stakes assessment: [Paste your performance task or major assessment] Key skills students need to practice: [List 2-3 skills the assessment requires] Course timeline: [How many weeks before the major assessment?] Please suggest 3-4 low-stakes practice activities that: 1. Target the same skills required by the major assessment 2. Can be graded on completion/effort rather than quality 3. Build progressively in complexity toward the final task 4. Include built-in feedback opportunities (self-assessment, peer review, or instructor check-ins) For each activity, specify: - Brief description and format - When it should occur relative to the major assessment - Suggested grading approach (completion-based, revision opportunity, ungraded) - How it scaffolds toward the high-stakes task
Dexi Says:
These add-ons are optional—use them if they fit your context. The assessment justification prompt is especially useful if you're introducing a new or unfamiliar assessment format. The low-stakes practice prompt works well when your performance task requires skills students haven't practiced before.
Step-by-Step AI Assessment Guide
This section provides detailed, step-by-step guidance on using AI to develop different types of assessments. Choose the assessment type you want to create and follow the process.
🎭 Developing Performance Tasks with AI
📝 Developing Traditional Assessments with AI
Designing AI-Resistant Assessments
As AI tools become more sophisticated, educators face new challenges in ensuring assessment integrity. Rather than viewing AI as only a threat, we can design assessments that either cannot be easily completed by AI or that meaningfully incorporate AI use while still demonstrating student learning.
Key Principle
The goal isn't to "catch" students using AI, but to design assessments where AI assistance either isn't useful or becomes a legitimate learning tool. Focus on assessing what matters: can students think, apply, and create?
Strategies for AI-Resistant Assessment Design
Synchronous Demonstration
Require real-time performance—oral exams, live presentations, in-class writing, or demonstrations where students must respond without AI assistance.
Personal Connection
Require students to connect content to their own experiences, local contexts, or class discussions that AI cannot access.
Process Documentation
Assess the learning process through drafts, reflection journals, or documented iterations rather than just final products.
Authentic Performance
Design tasks that require physical skills, human interaction, or judgment that AI cannot replicate.
AI-Resistant Assessment Examples
💡 Hover or tap each card to see the full description and why it works
Oral Examinations / Viva Voce
🔄 Flip for detailsOral Examinations / Viva Voce
One-on-one or small group verbal exams where students explain, defend, and extend their understanding in real-time.
Why It Works
Requires spontaneous thinking, follow-up questions probe depth, and reveals genuine understanding vs. memorization.
Live Problem-Solving Sessions
🔄 Flip for detailsLive Problem-Solving Sessions
Students solve novel problems in class while thinking aloud, explaining their reasoning as they work.
Why It Works
Assesses process and reasoning, not just answers. Instructor can ask clarifying questions in the moment.
Portfolio Defense
🔄 Flip for detailsPortfolio Defense
Students compile work over time and then present and defend their portfolio in person.
Why It Works
Even if AI helped create artifacts, students must explain their choices and demonstrate understanding verbally.
Observed Skills Assessment
🔄 Flip for detailsObserved Skills Assessment
Direct observation of clinical, laboratory, technical, or artistic skills performed in person.
Why It Works
Physical performance and procedural skills cannot be delegated to AI.
In-Class Writing with Prompts
🔄 Flip for detailsIn-Class Writing with Prompts
Timed writing in class responding to prompts revealed at the start. Can be handwritten or on locked-down computers.
Why It Works
Controlled environment, novel prompts, time pressure—all make AI use impractical.
Application to Class-Specific Content
🔄 Flip for detailsApplication to Class-Specific Content
Require students to apply concepts to case studies, examples, or discussions unique to your specific class sessions.
Why It Works
AI doesn't have access to what happened in YOUR class, requiring students to draw on their attendance and notes.
Video Explanations/Demonstrations
🔄 Flip for detailsVideo Explanations/Demonstrations
Students record themselves explaining concepts, solving problems, or demonstrating skills on video.
Why It Works
Requires personal appearance and verbal explanation that reveals authentic understanding.
Peer Assessment Activities
🔄 Flip for detailsPeer Assessment Activities
Students assess peers' work using rubrics, providing written feedback that demonstrates their own understanding.
Why It Works
Requires understanding to evaluate others; cross-references catch inconsistencies.
Iterative Assignments with Reflection
🔄 Flip for detailsIterative Assignments with Reflection
Multi-draft assignments where students submit iterations, explain changes, and reflect on their learning process.
Why It Works
Documenting the process of improvement is hard to fake; reveals authentic engagement with feedback.
Proctored Digital Exams
🔄 Flip for detailsProctored Digital Exams
Online exams with lockdown browsers, randomized questions, and time limits that make AI use impractical.
Why It Works
Technical barriers combined with time pressure and question randomization reduce AI feasibility.
Design an AI-Resistant Assessment: Step by Step
Sequential Prompting: Assessment First, Then Rubric
The prompts below separate assessment design from rubric creation. This produces better results than asking for both at once—you can focus on getting the assessment right before building evaluation criteria for it.
Help me redesign an assessment to be more AI-resistant. Current Assessment: [Describe your current assessment] Course: [Course name] Class Size: [Small/Medium/Large] Learning Outcomes: [What should the assessment measure?] Constraints: [Time, technology, TA support available] Please suggest: 1. Three Alternative Assessment Designs - Description of each approach - How it maintains the same learning outcomes - What makes it resistant to AI misuse - Practical implementation steps - Estimated time/resource requirements - Potential challenges and solutions 2. Hybrid Approach - How to combine AI-assisted preparation with AI-resistant demonstration of learning 3. Academic Integrity Statement - Clear policy language for the syllabus about AI use Do NOT include rubric criteria yet—I will develop those separately after selecting my assessment design.
Pause & Review
Review the three suggested designs. Which one best fits your learning outcomes, class size, and constraints? Is it genuinely AI-resistant, or could a student still use AI to complete it? Would your students find it meaningful? Select and refine one design before proceeding to the rubric.
Now I need a rubric for the AI-resistant assessment I've selected. Assessment description: [Paste or describe the assessment design you selected from Step 1] Learning outcomes assessed: [Paste the relevant learning outcomes] Please create an analytic rubric with: 1. 4-5 criteria aligned with the learning outcomes 2. 4 performance levels (Exemplary, Proficient, Developing, Beginning) 3. Specific, distinct, and observable descriptors (avoid vague terms like "appropriate" or "good") 4. Clear distinctions between adjacent levels 5. Suggested point values or weighting Format as a table with criteria in rows and levels in columns.
Pause & Review
Check each rubric descriptor: Is it truly specific and observable? Could a student read this and know exactly what to aim for? Do the levels represent meaningful distinctions? Revise any vague language before proceeding.
Review my assessment against my learning outcomes for alignment. My learning outcomes: [Paste all your course learning outcomes] My AI-resistant assessment: [Paste your finalized assessment description] My rubric criteria: [Paste your rubric criteria] Please check: 1. Are there any learning outcomes not adequately assessed by this assessment? 2. Are there any rubric criteria that don't connect to a stated learning outcome? 3. Does the assessment require students to demonstrate the cognitive level specified in the outcomes (e.g., if the outcome says "evaluate," does the task actually require evaluation)? 4. Suggest any adjustments to improve alignment.
Remember: It's About Learning
The best AI-resistant assessments aren't just harder to cheat on—they're better assessments. They focus on demonstrating genuine understanding, application, and the ability to think critically in ways that matter for students' futures.
AI-Integrated Assessments
Rather than viewing AI as a threat to academic integrity, we can design assessments that intentionally incorporate AI as a tool. These assessments teach students to work effectively with AI while still demonstrating their own learning, critical thinking, and professional judgment.
Why AI-Integrated Assessments?
Students will use AI in their careers. Teaching them to use AI effectively, critically evaluate AI outputs, and combine AI capabilities with human judgment prepares them for real-world success. The key is assessing the human skills that matter.
Types of AI-Integrated Assessments
AI as Research Assistant
Students use AI to gather information, then critically evaluate, synthesize, and apply it. Assessment focuses on analysis and application.
AI as First Draft Generator
Students prompt AI for initial content, then significantly revise, improve, and personalize. Assessment focuses on revision quality and critical editing.
AI Output Evaluation
Students evaluate AI-generated content for accuracy, bias, completeness, and quality. Assessment focuses on critical analysis skills.
AI as Thinking Partner
Students use AI to brainstorm, explore perspectives, or stress-test ideas. Assessment focuses on the quality of their final reasoning.
AI-Integrated Assessment Examples
💡 Hover or tap each card to see the full description and what you assess
AI-Assisted Literature Review
🔄 Flip for detailsAI-Assisted Literature Review
Students use AI to identify sources and summarize articles, then write a critical synthesis that evaluates the AI's suggestions and adds their own analysis.
What You Assess
Critical evaluation of sources, synthesis across articles, identification of gaps, and original analytical voice.
AI Fact-Checking Exercise
🔄 Flip for detailsAI Fact-Checking Exercise
Give students AI-generated content on a topic. They must identify errors, verify claims, and produce a corrected version with citations.
What You Assess
Ability to detect misinformation, research skills, attention to detail, and understanding of accurate content.
Prompt Engineering Challenge
🔄 Flip for detailsPrompt Engineering Challenge
Students must craft effective prompts to get AI to produce specific outputs, then document and reflect on their prompting strategies.
What You Assess
Understanding of AI capabilities, clear communication, iterative problem-solving, and metacognitive reflection.
AI-Enhanced Case Analysis
🔄 Flip for detailsAI-Enhanced Case Analysis
Students use AI to explore multiple perspectives on a case, then write an analysis that integrates, critiques, and extends beyond the AI's suggestions.
What You Assess
Critical thinking, perspective-taking, original analysis, and ability to build on AI-generated ideas.
Revision and Improvement Portfolio
🔄 Flip for detailsRevision and Improvement Portfolio
Students submit: (1) AI-generated first draft, (2) their significantly revised version, (3) a reflection explaining all changes and improvements made.
What You Assess
Editing skills, subject matter expertise applied to improvements, critical judgment, and metacognitive awareness.
AI Comparison Analysis
🔄 Flip for detailsAI Comparison Analysis
Students prompt multiple AI tools with the same question, then write a comparative analysis of the responses, evaluating quality, accuracy, and biases.
What You Assess
Analytical comparison skills, understanding of AI limitations, evaluation criteria development.
AI-Assisted Problem Solving
🔄 Flip for detailsAI-Assisted Problem Solving
Students document their problem-solving process using AI as a tool—showing prompts used, AI responses, and their own decision-making throughout.
What You Assess
Process documentation, strategic AI use, decision-making rationale, and problem-solving approach.
Human + AI Collaborative Project
🔄 Flip for detailsHuman + AI Collaborative Project
Students complete a project explicitly using AI for designated components (e.g., data analysis, drafting) while doing other components independently (e.g., research design, interpretation).
What You Assess
Strategic task allocation, quality of human contributions, integration of AI and human work.
Key Design Principles
Make AI Use Transparent: A Practical Guide
When students use AI as part of an assessment, require them to document their process. This transparency serves two purposes: it ensures academic integrity, and it develops metacognitive skills that are valuable in any professional context. Here are concrete formats you can use:
Prompt Log: Students submit the exact prompts they used, along with the AI's responses. This shows their ability to craft effective prompts and iterate on results.
Reflection Annotations: Students annotate the AI output, marking what they kept, what they changed, and why. This makes their critical thinking visible.
Before/After Comparison: Students submit their own draft alongside the AI-assisted version, with a brief explanation of how AI improved (or didn't improve) their work.
Process Narrative: A short paragraph describing how and why AI was used, what was helpful, what wasn't, and what the student would do differently next time.
Choose the format that fits your assessment. For major projects, a prompt log plus reflection works well. For shorter assignments, a brief process narrative may suffice. The key is making the human contribution visible and assessable.
Assess Human Value-Add
Design rubrics that specifically assess what the student contributed beyond AI output: critical analysis, contextualization, original insights, professional judgment, and quality improvements.
Design an AI-Integrated Assessment: Step by Step
Sequential Prompting: Assessment First, Then Rubric
As with AI-resistant assessments, these prompts separate the assessment design from the rubric. Get the assessment structure right first, then build evaluation criteria that focus on what the human contributes beyond AI output.
Help me design an assessment where students work WITH AI as a tool. Course: [Course name] Learning Outcomes: [What should students demonstrate?] AI Role: [Research assistant/First draft generator/Thinking partner/Other] Class Size: [Number of students] Please provide: 1. Assessment Description - What students will do - How AI is incorporated - What students must do independently 2. Transparency Requirements - How students document AI use - What they must submit (prompts, outputs, reflections) 3. Academic Integrity Guidelines - Clear policy for this specific assignment - What constitutes appropriate vs. inappropriate AI use 4. Student Instructions - Step-by-step process - Examples of good AI integration Do NOT include rubric criteria yet—I will develop those separately after refining the assessment design.
Pause & Review
Review the assessment design. Is the boundary between AI-assisted and independent work clear? Will students know exactly what AI can and cannot be used for? Is the transparency requirement practical—not so burdensome that it overshadows the learning? Refine the design before building the rubric.
Now I need a rubric for my AI-integrated assessment. Assessment description: [Paste your finalized assessment design from Step 1] Learning outcomes assessed: [Paste the relevant learning outcomes] Please create an analytic rubric with 4-5 criteria and 4 performance levels (Exemplary, Proficient, Developing, Beginning). The rubric must specifically assess what the STUDENT contributes beyond AI output. Include criteria such as: - Critical evaluation of AI output (identifying errors, gaps, or bias) - Quality of human improvements and original analysis - Effective AI prompting and iteration - Reflection on the AI collaboration process - Content accuracy and depth of understanding Use specific, distinct, and observable descriptors. Format as a table.
Pause & Review
Check the rubric: Does it genuinely assess human contribution, or could a student get a high score by simply presenting polished AI output? Are the criteria distinct enough that you could reliably distinguish between student work at different levels? Revise any descriptors that don't capture what quality human-AI collaboration looks like in your discipline.
Review my AI-integrated assessment against my learning outcomes for alignment. My learning outcomes: [Paste all your course learning outcomes] My AI-integrated assessment: [Paste your finalized assessment description] My rubric criteria: [Paste your rubric criteria] Please check: 1. Are there any learning outcomes not adequately assessed by this assessment? 2. Are there any rubric criteria that don't connect to a stated learning outcome? 3. Does the assessment require students to demonstrate the cognitive level specified in the outcomes (e.g., if the outcome says "evaluate," does the task actually require evaluation)? 4. Are the AI-related criteria (prompting quality, critical evaluation of output) aligned with outcomes about AI literacy or critical thinking? 5. Suggest any adjustments to improve alignment.
Generate reflection questions for students who used AI in their assignment. Assignment Type: [Description of the assignment] AI Tasks: [What students used AI for] Create 8-10 reflection questions covering: 1. Process Documentation - What prompts did you use? How did you refine them? - What worked well? What didn't? 2. Critical Evaluation - How did you evaluate the AI's output? - What errors or limitations did you identify? 3. Human Contribution - What did YOU add that the AI couldn't? - How did you improve upon the AI's work? 4. Learning Insights - What did you learn about working with AI? - How might you use AI differently next time? 5. Professional Application - How might this skill transfer to your career? - What are the ethical considerations?
AI-Assisted Activities
Complete these activities using your chosen AI tool. Remember the Human–AI–Human pattern: draft your thinking first, use AI to refine and expand, then critically evaluate the results.
Why Sequential Prompting Works Better
Each activity below uses two or more short prompts instead of one large prompt. This mirrors how expert designers actually work: generate a draft, review it, then build the next piece based on what you've refined. Sequential, conversational prompting gives AI better context at each step and gives you a checkpoint to catch problems early—before they compound into later outputs.
Return to DesignLab After Each AI Interaction
When you copy a prompt and paste it into your AI tool, remember to come back here afterward! Copy the AI's response, return to this page, and continue with the next step. AI chatbots can easily lead you down rabbit holes—DesignLab keeps your design process structured and focused.
Design a Performance Task
Step 1: Reflect
Before writing anything, think about your assessment goals: What realistic scenario or role could students take on? What product or performance would demonstrate their understanding? Which of your learning outcomes should this task assess?
Step 2: Rough Draft
Sketch out the basics of your authentic assessment: a brief scenario, the type of product students would create, and the outcomes it would measure. Even rough notes give you a strong baseline for evaluating the AI’s suggestions.
What Makes a Performance Task Strong?
Compare these two tasks for an Environmental Science course:
Weak: “Write a report on water pollution in your area.”
This is vague—no role, no audience, no constraints, no justification required. A student could summarize a Wikipedia article and technically complete the task.
Strong: “As an environmental consultant hired by your city council, analyze water quality data from three local sites collected over the past year. You have a $50,000 budget and 6 months to propose an intervention. Write a 5-page recommendation report for council members (non-scientists) that identifies the most urgent contamination source, justifies your recommended intervention over at least one alternative, and addresses likely community objections.”
The strong version has a discipline-specific role, realistic constraints, a requirement to justify a position, and a product that serves a real audience purpose. The prompt below helps you build tasks like this.
Help me design an authentic performance task for my [Course Name] course. Learning outcomes to assess: [Paste 2-3 learning outcomes from Module 2] Course context: [Level, student background] Using the GRASPS framework, suggest specific details for each element: - Goal: What is the student trying to accomplish? - Role: What discipline-specific role does the student take on? (e.g., consultant, researcher, practitioner—not just "student") - Audience: Who is the authentic audience, and what do they need from the product? - Situation: What is the realistic scenario, including constraints such as time limits, budget, incomplete data, or competing priorities? - Product: What will students create or perform? The product should serve a clear purpose for the audience. - Standards: What criteria define success? Additionally, the task must: - Require students to justify a claim, recommendation, or position (not just describe or summarize) - Include at least one realistic constraint (time, resources, uncertainty, or ethical tension) - Be completable in [timeframe] Discipline: [Your discipline/field—this helps AI generate field-appropriate scenarios] Please present each GRASPS element with 2-3 options so I can choose what fits my course.
Pause & Review
Before proceeding, review the GRASPS elements AI generated. Select the best options for each element and adjust anything that doesn't fit your discipline or students. Is the role realistic? Is the audience authentic? Once you're satisfied with the framework, use the follow-up prompt to generate the full task description.
Based on the GRASPS elements we just developed, please write the complete performance task description as students would see it. Include: 1. A clear, engaging scenario with the discipline-specific role and situation 2. Specific requirements for the deliverable and its purpose for the audience 3. Realistic constraints (time, resources, data limitations, or other boundaries) 4. A clear requirement for students to justify, defend, or argue a position 5. How the task connects to the learning outcomes Write in a professional but student-friendly tone.
Step 4: Critically Evaluate
Compare the AI’s task design with your rough draft. Is the scenario realistic for your discipline? Would your students find it meaningful? Does it actually measure the learning outcomes you intended? Revise until it reflects your expert knowledge of your field and your students.
Generate a Rubric
Step 1: Reflect
Think about what quality looks like for your performance task. What does excellent work look like in your discipline? What are the most important dimensions to evaluate?
Step 2: Rough Draft
List the 3–5 most important criteria you would use to evaluate student work on this performance task. Jot a brief note about what each criterion looks like at its best.
I need to build a rubric for this performance task: [Paste your performance task from Activity 1] Learning outcomes assessed: [Paste the relevant learning outcomes] Please suggest 4-5 rubric criteria that: 1. Are aligned with the learning outcomes and the standards for high-quality performance identified in the GRASPS task 2. Each capture a distinct, assessable dimension of quality 3. Are written in student-friendly language For each criterion, briefly explain why it matters and what aspect of the task it evaluates. Do NOT generate full descriptors yet—just the criteria and their rationale.
Pause & Review
Compare the AI's suggested criteria with the ones you drafted in Step 2. Are these the dimensions that matter most in your discipline? Are any critical criteria missing, or any suggested criteria unnecessary? Adjust, add, or remove criteria before generating the full rubric. Getting the criteria right first prevents wasted effort on descriptors you'll have to rewrite.
Using the criteria we confirmed, please generate a complete rubric. Confirmed criteria: [List your final criteria, adjusted from Step A] For each criterion, create 4 performance levels (Exemplary, Proficient, Developing, Beginning) with: - Specific, distinct, and observable descriptors (not vague terms like "appropriate" or "good") - Clear distinctions between adjacent levels - Language that tells students exactly what to aim for Format as a table with criteria in rows and levels in columns. After the rubric, suggest how to weight each criterion.
Step 4: Critically Evaluate
Compare the AI's rubric criteria with the ones you identified. Are the descriptors truly specific and observable, or do they use vague terms like "appropriate" or "good"? Do the performance levels represent meaningful distinctions in your discipline? Revise any descriptors that don't match what quality work actually looks like in your field.
Design Supporting Assessments
Step 1: Reflect
Look at your list of learning outcomes and the authentic assessment(s) you designed in the previous sections. Ask yourself:
• Which learning outcomes are already well-assessed by your performance task(s)?
• Which outcomes still need additional evidence—either because they aren't covered by the performance task, or because you want to check foundational knowledge along the way?
• For each outcome that needs additional assessment, what type would be most appropriate—a quiz, a formative check, a homework assignment, a discussion?
Not every outcome needs both a performance-based and a supporting assessment. Some may be fully addressed by your performance task alone.
Step 2: Rough Draft — Map Your Outcomes to Assessment Types
Create a quick mapping of your learning outcomes to assessment types. For each outcome, note how it will be assessed:
• Outcome 1: Performance task only / Supporting assessment only / Both
• Outcome 2: Performance task only / Supporting assessment only / Both
• Outcome 3: (and so on...)
For any outcome that needs a supporting assessment, jot down what format might work best (quiz, formative check, homework, discussion, etc.) and roughly when in the course it should occur. This map will feed directly into your AI prompt.
I am designing supporting assessments for my course. I have already designed authentic assessments, so I need to fill in the gaps with smaller assessments that check progress along the way. Course: [Course Name] Duration: [e.g., 15 weeks] Main performance task: [Brief description] Here is my outcome-to-assessment mapping. I have noted which outcomes are already covered by my performance task and which still need supporting assessments: [Paste your mapping from Step 2, e.g.: - Outcome 1: "Analyze X" — Covered by performance task; no additional assessment needed - Outcome 2: "Explain Y" — Needs a supporting assessment (foundational knowledge check) - Outcome 3: "Apply Z" — Covered by performance task, but would benefit from a mid-point progress check - Outcome 4: "Identify W" — Needs a supporting assessment (formative check)] Based on this mapping, please suggest supporting assessments ONLY for the outcomes that need them. For each suggested assessment, provide: - Which outcome(s) it addresses - Suggested format (quiz, formative check, homework, discussion, etc.) - When in the course it should occur - How it scaffolds toward the performance task Do NOT suggest assessments for outcomes that are already well-covered by the performance task. Do NOT develop detailed specifications yet—just the overall plan so I can decide which assessments to keep.
Pause & Review
Review the suggested assessment plan. Which of these assessments genuinely build toward your performance task? Are the timings realistic within your course schedule? Remove any that feel redundant, adjust formats that don't fit your teaching context, and confirm the assessments you want to develop further.
From the assessment plan, I'd like to develop these specific assessments in detail:
[List the 2-3 assessments you selected from Step A]
For each one, please provide:
1. Clear instructions as students would see them
2. Specific questions, tasks, or prompts included
3. Grading approach (points, completion-based, rubric)
4. How it scaffolds toward the major performance task
5. Estimated time to complete
Step 4: Critically Evaluate
Compare the AI’s suggestions with your rough draft. Do these supporting assessments genuinely build toward your performance task? Are the suggested timings realistic within your course schedule? Do the formats fit your teaching context and student needs? Adjust the assessment plan based on your knowledge of what works in your classroom.
Dexi Says:
AI-generated rubrics are starting points! Review each descriptor—does it match what you actually want to see? Adjust the language to fit your context and students.
👩🏫 Exemplar Spotlight: Stage 2 in Action
Here's how Dr. Sarah Chen and Prof. Marcus Rivera designed their assessments. Notice the choices they made—and the AI suggestions they rejected.
Dr. Sarah Chen — Assessment Design Decisions
Performance Task (GRASPS): Students adopt the role of a nurse on a medical-surgical unit who receives a new patient with multiple chronic conditions and a language barrier. They must conduct a comprehensive assessment, develop a prioritized care plan, communicate it to the interprofessional team (via written handoff report), and justify their clinical decisions in an oral debrief. Realistic constraints include limited time, incomplete chart information, and a family member who disagrees with the treatment approach.
Supporting Assessments: Weekly clinical reasoning journals (formative, completion-graded), medication safety quizzes (low-stakes knowledge checks), and a mid-term therapeutic communication role-play with peer feedback.
What Sarah Changed: AI initially proposed a straightforward case study with a single "correct" care plan. Sarah added ambiguity—the incomplete chart and family conflict—because real nursing practice is messy. She also rejected AI's suggestion for a written final exam, replacing it with the oral debrief because her accreditation standards emphasize clinical reasoning, not written recall. For the rubric, AI used vague descriptors like "appropriate nursing interventions." Sarah rewrote these to specify: "Identifies at least three priority nursing diagnoses and links each to a specific assessment finding."
Prof. Marcus Rivera — Assessment Design Decisions
Performance Task (GRASPS): Students work in pairs to investigate a specific form of structural inequality in their local community (e.g., housing segregation, food deserts, educational funding disparities). They produce a public-facing policy brief for a community organization that includes: structural analysis using at least two sociological frameworks, original data from community sources, and evidence-based policy recommendations. The brief must be written for a non-academic audience.
Supporting Assessments: Annotated bibliography building toward the policy brief (scaffolded in three stages), weekly discussion posts applying course theories to current events, and a reflexive positionality statement (completion-graded) exploring how their own social location shapes their analysis.
What Marcus Changed: AI suggested a traditional research paper. Marcus transformed it into a community-facing policy brief because he wants students to practice communicating sociological analysis to people who can act on it. He kept the requirement for scholarly sources but added the "non-academic audience" constraint to push students beyond academic jargon. For his rubric, AI generated criteria focused only on content accuracy. Marcus added criteria for community relevance and reflexive awareness—dimensions the AI hadn't considered.
Module 3 Reflection & Checklist
✅ Module 3 Completion Checklist
- Mapped which outcomes need performance-based vs. supporting assessment
- Designed at least one performance-based/authentic assessment
- Planned supporting assessments that scaffold toward performance tasks
- Created a rubric for my authentic assessment
- Verified alignment between outcomes and assessments
- Considered AI-resistant and AI-integrated assessment strategies
- Saved my work to the workbook
Your Reflection
📖 Further Reading for This Module
Want to go deeper into the frameworks and concepts introduced this module? These scholarly sources provide the foundation for the ideas in this module.
Wiggins, G. P., & McTighe, J. (2005). Understanding by design. Ascd.
CAST. (2018). Universal design for learning guidelines version 2.2. CAST.
🎉 Module 3 Complete!
Outstanding! You've designed assessments that will truly reveal student understanding.
Coming up in Module 4: Now we plan the learning experiences and instruction (Stage 3) that will prepare students for success!