A checkpoint plan for one term: four or five points, each tied to an assignment you already give, with a behavior focus and the evidence students turn in.
In Exercise 2, you built one assignment around one behavior. One assignment shows you a single moment. The BE behaviors develop over a whole term, so to see that growth, you check in a few times at points you plan ahead. Each of those check-ins is a checkpoint.
A checkpoint is usually a few questions added to work students already turn in, like a short note at the end of a paper, three questions on a Blackboard submission, or a brief journal entry. You don't need to create anything new. What students write there is your evidence. Grading from that evidence is fairer and more consistent than grading from a general impression of how a student seems to be using AI.
Keep the weight low
The college framework treats the BE behaviors as something students develop, and they're graded only when a course writes them into a student learning outcome. Even then, keep the weight modest, around 5 to 10 percent of the course grade. Early in the term, your feedback does most of the work. The grade can come at the end, once students have had time to practice.
Protect honesty
These rules keep checkpoints from working against the behaviors they're meant to build:
- Never take points off for disclosing AI use. If being honest costs points, students stop being honest. If a student used AI in a way your policy doesn't allow, handle that in a separate conversation, outside the checkpoint grade.
- Grade reflections on how specific and honest they are. A student who writes clearly about a time AI made their work worse has done the reflection well. There's no "right" conclusion to reach.
- Give students who don't use AI full credit for explaining their choice. Deciding not to use AI, and being able to say why, is good judgment.
- Use a scenario for BE Safe. You usually can't see what a student put into a tool, so ask what they would share, check, or keep out of AI in a situation from your field.
A sample plan for a 16-week term
| When | Checkpoint | What students do | Focus |
|---|---|---|---|
| Week 2 | Starting point | A short self-check, plus how they currently use AI or why they don't. Ungraded or completion only. | All five, lightly |
| Weeks 5–6 | First real assignment | A short note on whether and how they used AI, what they checked, and what parts are their own. | Honest, Critical |
| Weeks 9–10 | Midterm check-in | One time AI helped their learning, one time it didn't, and what they'll change. | Reflective, Responsible |
| Weeks 13–14 | Field or scenario check | A situation from your field: what they would share with AI, what they'd check, and what they wouldn't use it for. | Safe, plus judgment in the field |
| Finals | Portfolio | Three pieces of evidence from the term plus a short reflection on how their thinking changed. Graded holistically. | All five |
What makes a plan work
- Attach each checkpoint to an assignment students already do. In Blackboard, that can be a few questions on the submission or a short journal.
- Focus each checkpoint on one or two behaviors. Five at once is too many to give useful feedback on.
- Don't skip the Week 2 starting point. It's what lets the portfolio show growth, for students and for you.
- Keep the questions the same from checkpoint to checkpoint where you can, so students can see their own answers change.
Three ways to grade it
Process points and a portfolio (recommended). Give completion points and feedback at each checkpoint during the term, then grade the final portfolio holistically. This keeps the stakes low while students are practicing and puts the grade where there's the most evidence.
An evidence-based rubric. If you want more structure, use one row per behavior, with the playbook stages as the levels. Rate each row from something the student turned in, like a disclosure note, a checking note, or a reflection. Here's a sample row:
| Behavior | Introduction | Exploration | Application |
|---|---|---|---|
| BE Critical | Uses AI output as given | Checks some claims, but not consistently | Checks against course or field standards and explains what they changed or rejected |
To collect the evidence, add a short prompt to the assignment and ask students to turn in their answer with the work. Pick the one or two behaviors that fit the assignment. Every prompt works for students who don't use AI, too.
| Behavior | Prompt to add to the assignment | What a strong answer shows |
|---|---|---|
| BE Safe | "Which AI tool did you use on this assignment, if any? What do you know about what that tool does with what you type? What did you decide not to share with it, and why?" For students who didn't use AI: "Name one thing you would not share with an AI tool, and why." | Decides whether a tool fits the task before relying on it, protects other people's information, and anticipates where the tool is likely to fail in their field |
| BE Honest | "How did you use AI on this assignment? What did AI do, and what did you do? If you didn't use AI, say so." | Describes accurately what AI did and what they did, and asks when expectations aren't clear |
| BE Critical | "Choose one thing AI gave you. How did you check it? What did you change or leave out, and why?" | Checks against course or field standards and explains what they changed or rejected |
| BE Responsible | "Did you use AI on this assignment? Why or why not? Who could be affected by that choice?" | Chooses on purpose for each task, considers who's affected, and owns the result |
| BE Reflective | "Where did AI help your learning on this assignment, and where did it get in the way? What will you do differently next time?" | Changes how they work based on what they noticed, and can say what changed |
The last column is the Application end of each row. To turn any of these into a full rubric row, add what Introduction and Exploration look like in your course, the way the BE Critical sample does. The BE Safe prompt asks what students know and decided. It never asks them to show what they typed.
Rows on a professionalism rubric. Technical and health programs often already grade professionalism. Adding one or two BE rows, usually Honest and Responsible, fits the behaviors into a rubric students already know.
Work through the steps below, or skip to the Copilot prompt at the end and let it interview you. Your answers stay in this browser only. To keep a copy, choose Print or save as PDF when you finish the steps.
Step 1Your course
Step 2Plan your checkpoints
List four or five checkpoints. Start with the week, name the assignment it attaches to, choose one or two behaviors from the list, and say what students turn in. If you did Exercise 2, that assignment is a good candidate for one of these.
Behavior focus, checkpoint 1: Choose behaviors
Behavior focus, checkpoint 2: Choose behaviors
Behavior focus, checkpoint 3: Choose behaviors
Behavior focus, checkpoint 4: Choose behaviors
Behavior focus, checkpoint 5: Choose behaviors
Step 3Your starting point
Step 4Feedback and weight
Step 5Check your plan
Your answers save in this browser as you type.
Paste this into Copilot. It will interview you about your course and draft a checkpoint plan and the questions students answer. Don't paste in student names or student work, and read the draft carefully before you use it.
I'm a community college instructor planning a few checkpoints across one term to see how my students' judgment about AI develops. Each checkpoint adds a few questions to an assignment I already give. I'm not creating new assignments. The five behaviors: - BE Safe: knowing what goes into an AI tool and whether the tool fits the job. - BE Honest: being accurate about what you did and what AI did. - BE Critical: checking AI output before acting on it. - BE Responsible: choosing on purpose whether AI fits the task, and owning the result. - BE Reflective: noticing how using AI, or not using it, affects your learning, and adjusting. Interview me before you write anything. Ask one question at a time and wait for my answer. Keep it to about 6 questions. Find out: - My course, my students, and the length of the term - My main assignments and when they're due - Which behaviors matter most in my course or field - How I want to grade it: process points and a portfolio, a rubric, rows on a professionalism rubric, or feedback only Don't draft until I say "ready." Then write: 1. A plan of 4 or 5 checkpoints: when, which assignment it attaches to, one or two behaviors, and what students turn in. Start with an ungraded starting point early in the term and end with a short portfolio or final reflection. 2. The 2 or 3 questions students answer at each checkpoint, written to students in plain, warm language. Keep them short enough to add to a Blackboard submission. 3. A suggestion for how much it counts and when it's graded. Ground rules: - Students never lose points for honestly disclosing AI use. - Students who don't use AI can earn full credit by explaining their choice. - Grade reflections on how specific and honest they are, not on reaching a particular conclusion. - Assess BE Safe with a scenario from my field. - Keep the total weight low, around 5 to 10 percent. - Don't make up college policies. Flag anything that depends on OCTC or KCTCS policy for me to check. - Don't suggest AI detection tools. After the draft, ask me which checkpoint feels like the most work, and help me make it lighter.