A Guide to Relevant Assessments: Rethinking Performance Tasks in the AI Age
You’ve probably heard the term performance task before (especially if you are reading this article). If not, you’ve at least seen its cousins: the project, the portfolio, the presentation, or the dreaded poster board.
The performance task was born out of the desire to move beyond rote memorization and regurgitation tests, toward something that would measure deeper levels of learning. Grant Wiggins and Jay McTighe popularized this shift in their work on Understanding by Design, positioning performance tasks as authentic demonstrations of learning where students apply knowledge and skills to real-world problems.
I’ve written extensively about their work, and had McTighe on my podcast to share some of his research and perspective from year’s of shifting towards performance tasks.
The goal was always clear: move away from “telling” and toward “showing.” Students should be able to transfer their learning, not just regurgitate it.
But here’s the problem. In an age of artificial intelligence, where a chatbot can produce an essay, build a spreadsheet, design a slideshow, make a video, or even create a piece of art in seconds, how do we ensure performance tasks remain both relevant and authentic?
If we’re not careful, many of our performance tasks will turn into a different version of the “game of school”, with AI doing the heavy lifting, and students playing along.
So, how do we rethink performance tasks for this new reality?
Five Guiding Shifts for Relevant Assessments
Much like the Wheel of Responsibility framework gave us a new way to structure instruction, we need a fresh lens to view assessments. Here are five shifts worth exploring:
1. From Product to Process
In the past, performance tasks were often evaluated based on the final artifact: a research paper, presentation, or another creation. Now, the process becomes just as important as the end result.
AI can help students generate ideas, outline steps, and even draft portions of their work. But can they explain how they made choices? Can they reflect on revisions? Can they show evidence of thinking across multiple stages?
The assessment needs to capture the journey, not just the destination.
2. From Closed Prompts to Open Problems
A “tri-fold poster on Ancient Egypt” task doesn’t cut it anymore (it probably never did). AI can do the basics, fill in facts, and make it pretty.
Relevant assessments give students problems worth solving. And those problems are open-ended, messy, and complex. In our compliance-based system these types of tasks are often frowned upon. But imagine students solving one of these messy problems:
“How might we design a tourism campaign for Egypt that balances cultural preservation with economic growth?”
“What lessons from Ancient Egypt’s water management could be applied to modern cities facing drought?”
These problems don’t have a single answer. They demand creativity, discernment (I might need to write an entire post just on this idea), and real-world application. Each of these things AI can support, but not replace.
3. From AI-Resistant to AI-Integrated
Some educators instinctively look for tasks that AI “can’t do.” But if we only chase AI-proof prompts, we’ll miss the bigger opportunity.
In fact, I’ve argued many times (and written multiple articles) that we still need to be creating AI-Resistant learning experiences. However, there is no world where an assessment can be “relevant” if it doesn’t acknowledge the technology readily available in our times.
For this, we need AI-integrated performance tasks, where students are expected to use AI strategically, just as they would calculators, search engines, spreadsheets, or all kinds of other technology that we use in work, school, and life regularly.
The question shifts from: “Can AI do this?” to “How would/did you use AI, and why?”
4. From One-and-Done to Iterative Cycles
Real work doesn’t happen in one draft. Neither should student work.
Performance tasks in the AI age should build in multiple rounds: ideation, feedback, revision, reflection. When students share iterations of their work, we see how they respond to critique, how they refine ideas, and how they use tools (including AI) to get better.
This iterative process mirrors the world outside school, where “final” is almost never final.
Typically this doesn’t fit within the confines of a strict scope-and-sequence or bloated curriculum. We know the metacognitive piece is one of the most important aspects of learning, so let’s bring it back full force for today’s learners.
5. From Teacher-Owned Rubrics to Co-Created Criteria
Traditionally, rubrics have been written by teachers and handed down to students. In relevant assessments, criteria should be co-created.
When students help define what quality looks like, they take ownership of the work. And in an AI world, where polished products can be manufactured in minutes, student voice in setting expectations ensures the focus stays on meaning, not shortcuts.
As we wrote in our book EMPOWER, “What decisions am I making for students that they could be making for themselves?”
A Wheel of Relevant Assessment? An Inverted Bloom’s Taxonomy?
If instruction was never linear, then assessment isn’t either.
I think of relevant assessments less as a checklist and more as a wheel that keeps turning:
Ideate (generate ideas, draft, use AI support)
Create (build, design, prototype, perform)
Reflect (pause, analyze, self-assess)
Revise (improve based on feedback, iterate again)
Share (publish, present, or demonstrate to an authentic audience)
The wheel keeps moving. And as it does, both students and teachers stay engaged in the cycle of learning.
John Spencer and I wrote about this process in our book, LAUNCH, and even designed a Launch Cycle that takes this iteration to the next level.
Another way to think about it could be the Inverted Bloom’s Taxonomy.
Dr. M. Workmon Larsen recently wrote a fantastic piece around the changes we are seeing in education and learning with the advent of generative AI. I urge you to read the entire article here, but she gets to the real crux of this issue when she discusses how AI is now impacting what we believed to be true with Bloom’s Taxonomy.
Traditional Bloom’s Taxonomy is a climb — learners start at the bottom, remembering and understanding concepts, before advancing to applying, analyzing, evaluating, and eventually creating. This made sense in a world where knowledge was considered more static or even a smidge more linear, but AI disrupts this model.
The gateway of learning in the age of AI is creation.
Starting with creation encourages learners to experiment, build, and test real-world ideas, sparking curiosity and uncovering the questions that drive deeper understanding. By engaging in creation, learners explore how tools work through hands-on experimentation and real-world application.
Learners begin by evaluating their outcomes, asking: What needs to change to improve this result? They then move to analyzing the problem: Why did this approach lead to this outcome? Next, they apply their insights to adjust and refine their work: What can I do differently next time? Through this process, learners build understanding by connecting their discoveries to broader concepts and systems. Finally, they remember key lessons and integrate them into future iterations.
Create is the entry point, not the pinnacle. It drives learners to evaluate the outcomes of their work, analyze what worked (and what did not), and apply what they’ve learned in new ways. This isn’t about skipping foundational knowledge — it’s about embedding that knowledge in meaningful action.
Ideation and creation can now be the entry point. Learners defend their knowledge, decisions, and transfer their understanding and skills to new authentic applications.
This may scare you. It sure looks different.
But, things have changed. We can’t roll out the same playbook year after year. It’s time to adjust, as we always have in education, and change our assessments in the AI age.
The Kicker: Relevant Assessments are Authentic by Design
AI isn’t going away. Neither is the push for deeper learning.
Relevant assessments don’t pretend students live in a pre-AI world. Instead, they invite students to work with the tools at hand, wrestle with problems worth solving, and showcase thinking that AI can’t replicate.
The verbs in that sentence matter. Work, wrestle, showcase. An in a world filled with technology, to ban it from assessments means we are the opposite of relevant.
The standards may be the same, but there is nothing standardized about a performance task that is iterative, authentic, and deeply connected to the real world.
Relevant assessments lead to authentic problem-solving.
Authentic problem-solving leads to meaningful learning.
And in the end, that’s the point: not to play the game of school, but to prepare students for the game of life (for which they are already playing).
