Prompt Libraries Are Not Enough. Students Need a Record of Judgment.
A simple classroom tool for making student AI use visible, explainable, and worth discussing
This semester, in MGMT 3210 — Business Communication Theory and Practice, I started using a Group AI Use Log for student group projects. You can view, copy, and adapt the Group AI Use Log here.
I made it after noticing a gap in the way we talk about AI in higher education. There is no shortage of advice on prompting. Students can find prompt libraries, prompt templates, and prompt examples in minutes. What they are less likely to find is a practical way to account for how they used AI, what they took from it, what they rejected, and what they learned in the process.
That gap matters.
When students use AI in academic work, the important issue is rarely whether they can get polished text from a tool. The important issue is whether they can assess that output, decide what is useful, reject what is weak, and explain the difference. That is where judgment shows up. That is also where learning shows up.
The log was my attempt to make that part visible.
Prompt libraries stop too soon
Prompt libraries can be useful. They help students get started. They show how wording can shape output. They sometimes save time when a student is stuck.
They also stop too soon.
They focus on the request, not the response. They tell students how to ask for something, but not how to evaluate what comes back. They do not show whether the student checked the output, revised it, compared options, or accepted it simply because it sounded smooth.
That matters because polished language can cover weak thinking.
A student can paste a prompt into ChatGPT or Gemini and get back a clean paragraph, a tidy outline, or a list of recommendations that sound plausible. None of that tells you whether the content is accurate, useful, or worth keeping. None of it tells you whether the student exercised judgment.
That is the weak point in a lot of current AI guidance for students. It helps with output generation. It says much less about decision-making.
The idea came from a research log
Part of the idea came from my Qualitative Research PhD course. In that course, we maintain a research log as one of the assignments in our research portfolio. You record what you did, what you noticed, what changed, what held up, and what did not.
That practice does something useful. It creates a record of decisions. It forces you to slow down and notice your own process. It gives you something more helpful than a polished final product with no visible path behind it.
I wanted something similar for student AI use.
So I built a simple log for MGMT 3210 group work. The document asks students to record any meaningful use of AI that influenced thinking, drafting, analysis, revision, or decision-making for the assignment. The purpose is stated clearly. The goal is not to penalize AI use. The goal is to show judgment, process, and learning.
That framing changes the tone immediately. It tells students this is not only about disclosure. It is about accounting for their choices.
What students actually record
The log is not complicated.
For each meaningful use of AI, students record the date and approximate time, the stage of work, the tool used, the purpose of the AI use, one representative prompt or exchange, a summary of the output, what they used or rejected, and a short reflection on how the AI use affected their work or learning.
That structure moves students beyond a simple statement like “I used AI.”
That sentence tells you almost nothing.
A sentence like this tells you much more: “On March 18, during revision, I used Gemini to test three ways to tighten the opening of our report. I kept one structural suggestion, rewrote the wording myself, and dropped the rest because it made the tone too generic.”
That is useful. It shows purpose, output, choice, and revision.
The log also tells students not to record minor or routine uses unless those uses made a substantial change to the final submission. That one instruction keeps the whole exercise grounded. Students do not need to catalog every tiny interaction with a tool. They need to document the moments when the tool changed the work in a meaningful way.
Group work is where this gap becomes obvious
One thing that became clearer after I shared the log more broadly is how few student-facing AI resources are built for group assignments.
That gap is worth paying attention to. Group work is a standard part of academic life. It is also one of the places where AI use becomes hardest to see clearly.
One student uses Gemini to generate early ideas for a business report. Another uses ChatGPT to test a stronger structure for the analysis section. A third uses Copilot to tighten wording on presentation slides the night before the group presents. By the time the work is submitted, those decisions are spread across several people. The final product gives very little indication of what happened along the way.
The log deals with that directly. Groups submit one AI use log per deliverable, with separate logs for the business report and the presentation. Each time a group member makes meaningful use of AI for a deliverable, that group member completes a separate entry.
Get krishans’s stories in your inbox
Join Medium for free to get updates from this writer.
That is a practical design choice. It keeps the record tied to real contributions. It avoids the familiar group-work problem where process disappears behind a vague statement such as “we used AI a bit.”
A student comment that clarified the point
A student in the course told me that prompt libraries are easy to find, but the reflection process in this log felt like the missing part.
That comment stayed with me because it captured the problem cleanly.
Students already have access to many examples of prompts. What they do not often have is a structure that asks them to account for the choices they made after the prompt. That is where the educational value is.
Prompting is easy to teach because it is visible and concrete. Judgment is harder to teach because it shows up in revision, rejection, editing, and restraint.
That is where this log does its work.
The more common problem is uncritical use
There is a lot of concern about improper AI use in academic work, and some of that concern is warranted. In day-to-day teaching, another problem shows up just as often.
Students accept output too quickly.
They get a response that sounds clear and finished, and they move on. They do not stop to test the logic, check the fit, or ask whether the output made the work better. They confuse fluency with quality.
A reflection log creates useful friction at that exact point.
When students have to say what they kept, what they changed, and what they rejected, they have to inspect the output. When they have to reflect on whether the tool improved their understanding or mainly helped with efficiency, they have to separate learning from convenience.
That is a good discipline. It does not solve every problem. It does push students toward a more active stance.
Documentation still needs boundaries
The document also states something that should be obvious, but is still worth saying plainly: documenting AI use does not make every use appropriate. Students still have to follow assignment instructions and course rules about acceptable use.
That line matters.
A log should not become a loophole where students assume any use is acceptable as long as they mention it. Reflection matters, but it does not replace course expectations. Good documentation supports responsible use. It does not excuse poor choices.
There is another practical note in the document that came from actual classroom conditions at MRU. My students were part of the Gemini pilot, and the broader rollout begins in the Spring 2026 semester. Since we work in a Google Workspace environment, Gemini interactions are retained in that system. That made it unnecessary to require screenshots or full chat transcripts with every assignment. The log still needs to be accurate, and I may ask to see the relevant Gemini conversation if I need to verify an entry. That keeps the process accountable without turning every assignment into a file archive.
What faculty need is something usable
Most faculty do not need another abstract debate about AI. They need something they can use with students now.
They need a structure that is simple enough to assign, short enough to read, and clear enough that students know what is expected. They need something that deals with the work in front of them.
This log does that for me.
It came out of one course. It borrows from a research practice that has already been useful in another context. It gives me a better way to talk with students about process, accountability, and judgment.
That is enough reason to keep using it.
What students need more than prompts
Students already have enough encouragement to generate text quickly.
What they need is practice deciding what deserves to stay on the page.
They need practice spotting weak logic under polished language. They need practice explaining why they used a tool, what it changed, and what they refused to take from it. Those are ordinary academic habits. AI gives us a new place to ask for them.
That is why I think the missing piece is not another prompt formula. It is a record of judgment.
That is what this log is trying to provide.
Visual note: The images in this article were created with ChatGPT.
A Note on Reuse
View, copy, and adapt the Group AI Use Log
This Group AI Use Log is shared under a Creative Commons Attribution (CC BY 4.0) license. That means you’re free to reuse, adapt, and remix it — whether you’re an educator, student, or institution — so long as you give proper credit to me, Kris Hans. AI in education is evolving fast, and we all benefit from sharing what works (and what doesn’t).
If you do use this Group AI Use Log, I’d love to hear how it’s working for you. Feel free to reach out!
