Start With One Assignment, Not Your Entire Course

As I wrap up my series of posts, AI is Exposing Existing Problems, I find myself returning to a simple question: Now what?

Throughout the first series, I explored some of the challenges AI is exposing in higher education. We looked at assessment, student motivation, transparency, evidence of learning, visible thinking, and the evolving role of faculty.

But understanding the challenge is only the first step. The next question is what we actually do about it.

So, over the next eight posts, I want to shift from theory to practice. Rather than focusing on what AI is changing, I'll explore practical instructional design strategies that faculty can implement without redesigning an entire course.

We'll look at assignment design, reflection activities, discussion strategies, verification skills, transparency practices, and other approaches that can help keep student thinking visible in an AI-rich world.

And I'll start with a suggestion that may be less intimidating than many faculty expect: Don't start with your entire course. Start with one assignment.

One of the biggest mistakes I see in conversations about AI and teaching is the assumption that responding to AI requires redesigning an entire course. That's a daunting proposition. Faculty are already balancing course preparation, grading, student support, committee work, professional development, accreditation requirements, and countless other responsibilities. Adding "rebuild your entire course for AI" to that list can make the challenge feel overwhelming before it even begins. And honestly, I don't think it's necessary.

The good news is that meaningful instructional change rarely starts with a complete redesign. In fact, educational improvement has often been driven by small, intentional changes rather than sweeping transformations. Whether we're talking about active learning, formative assessment, universal design, or metacognitive practices, many of the most effective teaching strategies are often introduced one activity, one assignment, or one class session at a time.

The same may be true for AI. Rather than asking, "How do I redesign my entire course?" a more productive question might be: "Which assignment concerns me the most?"

Choose the assignment where you're least confident that the final product reflects actual learning. Choose the assignment where you're most concerned that students could complete the work without deeply engaging with the material. Choose the assignment where AI appears most capable of producing something that looks successful.

Then ask a simple question: "What evidence of learning am I actually looking for?"

I find that question shifts the conversation in important ways.

For years, many of us have designed assignments around the final artifact:

·       the paper

·       the presentation

·       the discussion post

·       the project deliverable

Those products still matter. But AI has exposed something that may have been true long before generative AI arrived. A completed assignment is not necessarily the same thing as evidence of learning.

Educational researcher John Biggs argued through his work on constructive alignment that assessments should provide evidence that students have achieved intended learning outcomes. The challenge AI presents is that it forces us to look more carefully at whether our current assignments are actually providing that evidence.

If a student submits a polished essay, what exactly are we learning from that essay? Do we know how they arrived at their conclusions? Do we know what decisions they made during the process? Do we know how they evaluated information, weighed alternatives, or developed their understanding?

Sometimes the answer is yes. But sometimes the answer is less clear than we would like. That doesn't mean the assignment is broken. It simply means there may be opportunities to strengthen the evidence of learning it provides.

Fortunately, those changes are often surprisingly small. A reflection prompt asking students what challenged their thinking. A justification question requiring them to explain why they selected a particular approach. A process checkpoint where they document key decisions before submitting a final product. A real-world application that requires them to connect course concepts to a novel situation.

None of these additions fundamentally change the assignment. Yet each creates opportunities to make student thinking more visible.

And that idea, visible thinking, may be one of the most important instructional design principles in an AI-rich environment.

Researchers in metacognition have long argued that learning improves when students are encouraged to examine their own thinking processes. Similarly, formative assessment research emphasizes the importance of making learning visible so that both students and instructors can better understand current levels of understanding.

AI does not diminish the importance of these ideas. If anything, it may increase it.

As AI tools become increasingly capable of generating polished outputs, the value of education may shift even further toward helping students develop judgment, reasoning, reflection, and self-awareness. Those capacities often reveal themselves not only in what students produce, but in how they produce it. This is one reason I encourage faculty to resist the urge to redesign everything at once.

Large-scale course redesign can be valuable, but it can also create paralysis. The scope feels so large that meaningful action never begins. Starting with one assignment feels different. It's manageable. It's measurable. And it creates an opportunity to experiment.

You can introduce a small change, observe how students respond, and decide what adjustments make sense moving forward. In many ways, this mirrors how learning itself works.

Students rarely transform overnight. They develop understanding incrementally through practice, reflection, feedback, and revision. Course improvement often follows the same pattern. Small adjustments accumulate over time. And those adjustments can have a surprisingly large impact.

The conversation around AI sometimes creates the impression that higher education stands at a crossroads requiring immediate and dramatic change. I understand that impulse. But I suspect many of the most successful responses will be far less dramatic.

They will begin with one assignment. One question. One opportunity to make student thinking a little more visible than it was before. Because course transformation rarely begins with a complete overhaul. It usually begins with one thoughtful change.

And once we identify an assignment we want to improve, the next question becomes: What kinds of assignments actually keep students thinking when AI is readily available?

Continuing the Conversation

Series 1: AI Is Exposing Existing Problems ✓ Completed

Series 2: What We Do About It

Current Post (1 of 8): Start with One Assignment, Not Your Entire Course

Next Up: Designing Assignments That Keep Students Thinking