From AI Rules to Learning Goals

Throughout this series, I've argued that AI is pushing us to ask better instructional design questions. We've talked about designing assignments that keep students thinking, making thinking visible, building reflection into learning, teaching verification skills, and creating discussions that promote judgment rather than information recall.

But there's one design element we haven't talked about enough.

How we explain our expectations.

When faculty ask me about AI policies, the conversation often begins with questions like:

·       Should students be allowed to use AI?

·       How much AI is too much?

·       Should students disclose AI use?

·       What belongs in the syllabus?

Those are important questions. But I increasingly think they're the second questions we should ask. The first question is much simpler: What is this assignment designed to help students learn?

Because once we can answer that question clearly, AI expectations become much easier to explain.

For years, educational researchers have emphasized that students learn more effectively when instructional activities, assessments, and learning outcomes are clearly connected. John Biggs described this through the idea of constructive alignment. More recently, researchers involved in the Transparency in Learning and Teaching (TILT) project have demonstrated that making the purpose of assignments explicit can improve student learning, confidence, and academic success.

In other words, transparency isn't simply about telling students what to do. It's about helping them understand why they're doing it.

That distinction matters.

Earlier in this series, I wrote about what I called the missing purpose problem. AI didn't create that problem. It exposed it. When students don't understand why they're completing an assignment, it's easy to see the work as another task to finish rather than an opportunity to develop knowledge or skills.

AI raises the stakes.

Students now have access to tools that can brainstorm ideas, organize outlines, explain concepts, generate drafts, and solve many of the tasks we've traditionally assigned. If students don't understand what learning an assignment is intended to produce, AI expectations can feel arbitrary.

·      "Don't use AI."

·      "Only use AI for brainstorming."

·      "Disclose all AI use."

Without context, those statements are simply rules. Rules may produce compliance. Understanding supports learning.

That is why I think every AI expectation should begin with the learning goal.

Instead of writing: "AI may not be used on this assignment."

We might write:

Learning Goal: The purpose of this assignment is to help you develop your ability to construct and organize your own argument using evidence from course materials.

AI Expectations: You may use AI to brainstorm ideas or ask questions about the topic. However, because developing your own reasoning is the primary learning goal, AI should not generate the argument or written response you submit.

Notice what changed. The restriction didn't really change. The explanation did.

Students now understand that the expectation isn't about distrusting AI. It's about protecting the cognitive work the assignment was designed to develop. The same principle applies when AI use is encouraged.

Suppose the goal of an assignment is evaluating competing solutions.

Instead of simply saying: "AI use is permitted."

We might write:

Learning Goal: This activity is designed to strengthen your ability to evaluate competing recommendations and justify your decisions.

AI Expectations: You are encouraged to use AI to generate possible solutions or alternative perspectives. Your grade, however, will be based on how effectively you evaluate those ideas, explain your reasoning, and support your conclusions with evidence.

Again, AI isn't the focus. Learning is.

One simple framework I've started thinking about is that every AI expectation should answer three questions:

·      What am I supposed to learn?

·      What role can AI appropriately play in that learning?

·      Why does that level of AI support the learning goal instead of replacing it?

When students understand all three, expectations become much easier to interpret because they are grounded in purpose rather than policy.

I also think this changes the relationship between students and instructors. When expectations are explained only as rules, conversations often become about compliance. "What happens if I use AI?" "Is this allowed?" "Will I get in trouble?"

But when expectations are connected to learning goals, the conversation becomes instructional. "What thinking am I supposed to do?" "What part of this process should belong to me?" "How can AI support my learning without replacing it?"

Those are much better educational questions.

Throughout this series, I've argued that AI is encouraging us to shift our attention away from producing information and toward developing judgment, reasoning, reflection, and visible thinking.

I think our AI expectations should reflect that same philosophy. Increasingly, I see AI transparency statements not as policy documents but as instructional design tools. Their purpose isn't simply to regulate technology.

Their purpose is to explain learning. And when students understand both the purpose of an assignment and the role AI should play within it, expectations become clearer, trust becomes easier to build, and AI becomes another tool that supports learning instead of replacing it.

Series 1: AI Is Exposing Existing Problems ✓ Completed

Series 2: What We Do About It

Current Post (7 of 8): From AI Rules to Learning Goals

Next Up: Designing for the Thinking We Can't Outsource