5 Levels of AI Use for Students
Table of Contents
TL;DR: A Structured Approach to Student AI Use
(Instructional Guidance, Transcript, and Prompts below)
Why a Structured Approach Matters
Rather than treating all student AI use as cheating, this framework provides teachers with a clear, instructional structure for defining when and how AI may be used while still protecting learning goals and academic integrity.
The Five Levels of Student AI Use
Level 1: No AI Use
Students complete the assignment entirely on their own, using only their own knowledge and skills. This level is best suited for assessing foundational skills, recall, and independent fluency.
Level 2: Brainstorming with AI
AI is used only to generate ideas, topics, or strategies. The student remains fully responsible for creating the final product independently.
Level 3: Drafting with AI
AI generates an outline or rough draft based on student direction. The student must significantly revise, refine, and take ownership of the final submission.
Level 4: Creating with AI
AI contributes portions of the final product under close student supervision. The student directs the work, evaluates quality and accuracy, revises outputs, and integrates them into a cohesive whole.
Level 5: AI as a Co-Worker
AI functions as a collaborative partner that independently produces content under student direction. The student remains accountable for accuracy, purpose, and learning, and must be able to justify how and why AI was used.
How to Choose the Appropriate Level
Assignment Purpose
Teachers should first clarify why the assignment exists and what learning it is intended to measure or develop.
Cognitive Demand
Lower-level tasks focused on recall align with lower AI levels, while higher-order tasks involving analysis, synthesis, and creativity can support higher levels of AI use.
Student Readiness
Students must be developmentally and instructionally ready to manage, direct, and evaluate AI use before engaging at higher levels.
Key Takeaway
The five-level framework gives educators a practical, flexible way to define acceptable AI use by assignment, shifting the conversation from punishment to purposeful instructional design.
Teacher Take-aways
Instructional Principles for Student AI Use
Teachers can treat AI permissions as a task-design decision aligned to instructional purpose, rather than as a blanket rule. The central distinction is between assessment (measuring what a student can do independently) and learning activities (supporting growth through practice, feedback, and revision). When the goal is to evaluate independent fluency, foundational skill, or recall, AI support can invalidate the evidence of student performance. When the goal is learning, AI can be used as a scaffold if the assignment design keeps the student cognitively responsible for the key thinking and decision-making (Memarian et al., 2024).
Protect the Cognitive Core of the Assignment
Effective AI integration begins by identifying the “cognitive core” (the non-negotiable thinking students must do to meet the objective). If the objective targets analysis, argumentation, synthesis, problem-solving, or explanation, then the assignment should require students to demonstrate those capacities directly, even if AI assists with supporting steps. Research on cognitive offloading indicates that people tend to shift work to external tools when tasks feel demanding, which can reduce internal cognitive effort unless the task is designed to require reflection and active control (Risko & Gilbert, 2016). Practically, this means teachers can allow AI to help generate options or drafts while still requiring students to make the major intellectual moves themselves.
Require Comprehension, Not Just Production
A strong pedagogical guardrail is to require students to understand and evaluate any AI-generated content they use. This can be operationalized through brief, targeted “understanding checks” embedded in the workflow, such as
-
Students explain key claims or steps in their own words;
-
Students justify why a particular AI suggestion was accepted, revised, or rejected; and
-
Students identify potential inaccuracies, bias, or missing perspectives and correct them.
These moves keep learning visible and reduce the risk that AI becomes a substitute for comprehension. Research on self-regulated learning supports designing activities that prompt students to plan, monitor, and evaluate their work, because these are processes that improve learning and transfer (Panadero, 2017).
Match AI Autonomy to Student Readiness
Students vary in their ability to direct tools, evaluate quality, and revise strategically. Studies on AI literacy and competency frameworks emphasize that productive AI use depends on knowledge, judgment, and reflective mindsets, not only tool access (Chiu, 2024). Instructionally, this means AI permissions can function like scaffolding: as students demonstrate stronger self-regulation and evaluative skill, teachers can increase autonomy. When readiness is lower, teachers can tighten guardrails and add structure (for example, required checkpoints, templates for prompts, or mandatory revision logs).
Teach AI Use as a Learnable Set of Academic Practices
A practical approach is to treat AI use as a set of academic practices that can be taught, modeled, and assessed, such as
-
Asking precise questions and setting constraints for outputs;
-
Checking claims against course texts and trustworthy sources;
-
Revising for accuracy, alignment to criteria, and clarity; and
-
Documenting decisions and changes across drafts.
Recent research reviews of large language models in education consistently emphasize both the opportunities for learning support and the need for explicit guidance to address limitations, reliability, and responsible use (Kasneci et al., 2023; Zhang et al., 2024).
References for Further Discussion
Chiu, T. K. F. (2024). What are artificial intelligence literacy and competency? A comprehensive framework for K–12 AI education. Computers and Education: Artificial Intelligence, 5, 100173. https://www.sciencedirect.com/science/article/pii/S2666557324000120
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103, 102274. https://www.sciencedirect.com/science/article/pii/S1041608023000195
Memarian, B., Schuck, S., & McDonald, M. (2024). A review of assessment for learning with artificial intelligence. International Journal of Educational Research Open, 6, 100353. https://www.sciencedirect.com/science/article/pii/S2949882123000403
Panadero, E. (2017). A review of self-regulated learning: Six models and four directions for research. Frontiers in Psychology, 8, 422. https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2017.00422/full
Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, 20(9), 676–688. https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613%2816%2930098-5
Zhang, P., Li, Y., & Wang, X. (2024). A systematic review of ChatGPT use in K–12 education. European Journal of Education, 59(3), 1–23. https://onlinelibrary.wiley.com/doi/10.1111/ejed.12599
Full Transcript
A Structured Approach to Student AI Use
What’s your approach to kids using AI in your classrooms? On one hand, we can say if kids use AI, they’re cheating and we’re going to fail them. Or we can provide a structure, a way to define what they’re allowed to do and yet still accomplish our learning purposes.
What we need is the five levels of AI use, and that’s what we’re going to talk about here.
So here are the five levels briefly all the way from Level one, which is “You may not use AI on this particular assignment,” all the way through using AI as a co-creator of our assignment as a co- creator of our products.
Let’s talk about each one of these, what they look like and what they’re used for.
Level 1 AI Use: No AI Use
Level one is the simple one: for this assignment, no AI. The student must use his or her own skills and knowledge and abilities.
And we’re going to use level this primarily when we’re trying to assess students for their foundational skills or anything that requires recall by the student. That’s level one.
Now, what does it look like in practice? Here’s a math assignment. Do it on your own. Here’s your research topic, and over there are your resources. Go and write your own paper. Or take these texts and give me a summary in your own words.
So that’s level one in practice: no AI.
Level 2 AI Use: Brainstorming with AI
Level two is where we’re actually starting to use AI, but at a very low level.
Level two is AI-assisted brainstorming. So, if we’re trying to generate ideas, generate information that the student will then use independently, that’s level two.
The AI is just an idea generator here, but the student is responsible for the final product.
So, if students have to write or do a project about a historical period, the AI can generate some ideas that the student can then use in the product. Or asking kids to come up with a topic for a science project is a frustrating experience, to say the least, right?
The AI can help generate possibilities for them or strategies for completing a product, but then the student does the work independently.
Level 3 AI Use: Drafting with AI
Level three is where the AI is actually beginning to contribute to the content of the product or content of the assignment.
The AI is generating a first draft, a rough draft, or generating an outline for it. Then the student takes that draft, significantly revises it, guides the drafting process, directs the AI exactly what should be done and what it should cover to produce that rough draft, which the student then takes, revises, and submits.
So a first draft for a research paper: that’s level three. Or if the student says, “Here all of my notes, take my notes, take my information, create a rough draft for me that I’ll work on,” or “I need an infographic about this topic. Create a mock up for me as a guide that I can use to do my own work.”
That’s all Level three.
Level 4 AI Use: Creating with AI
And Level four: AI Collaborative Creation.
The AI is now beginning to contribute content to the final product. Think about it like this. If the student is the boss, the AI is the intern. And the boss says, “Here’s what I want you to do, and here’s what it needs to look like, and here’s the structure” and so forth. “Go create part of it for me. Go do it, and then let’s take a look at it together. Revise it, incorporate it, make sure it meets the purpose and do additional revisions on it.”
So here, the student is really in charge of directing the AI’s work, reviewing the work, revising the work, and incorporating it into a larger document or larger product.
The AI is producing, but the student is firmly in control of the ideas and the direction and the structure and the nature of it.
Basically, we’re going to use Level four when we have analytical tasks, or we need to creatively synthesize information. We have the AI do it, but then the student is responsible for revising and incorporating into the larger body.
Certainly, the AI is not getting the grade for the assignment, so the student has to be able to demonstrate understanding and mastery of the content that’s being produced.
Now, in practice, it might look like this. The student has to give a presentation. The AI can actually write part of the script for that presentation, or the flow of how the presentation is going to go that the student then delivers.
Or the AI says, “Here is a synopsis of information,” and the student is going to take it, review it, make sure it’s appropriate, complete, comprehensive, etc., and fit it into the final product. Or any time we ask the AI to generate visuals based on the information that we give it, and then we put it into our report or our product, that’s level four.
Level 5: AI as a Co-Worker
Now, finally we get to level five. This is the highest use here.
So thinking about our boss motif, we don’t have a boss and an intern here. What we have two coworkers, with one, the student, being the lead team member. But they are both able to work independently and contribute meaningfully to the final product.
However, the caution is the student is still responsible for the product and making sure it’s appropriate and accurate and so forth. But the AI is actually producing parts or content independently with the direction of the student.
The student has to be able to justify why did they have the AI do that and what did the AI produce.
So we would use something like this when we have a capstone project. For example, you are taking a lot of different products we put together, and we’re having the AI take part of it and create something new for us.
Or anytime we need new ideas or new information that perhaps isn’t accessible to the student and then do something with it, that’s the AI serving as a team member and contributing meaningfully to the end.
But, as always, the student has to, the student is still responsible for knowing what it is and knowing the content, because if it’s an assignment (as opposed to an assessment), it’s an assignment, it’s part of the learning process. So even if the AI is producing something and contributing meaningfully, the student still needs to learn from what the AI produced.
In practice, the student might say. Here are seven key events I need information about. And the AI goes, learns about them and writes a synopsis of each of those. Or the AI is given a topic and goes out and does research and provides a summary of key findings or key information. So the AI is out there doing the work, which the student then incorporates.
Anytime we’ve got the AI creating formulas or creating processes, then executing them, and then determining what the results are—that’s level five. Or if the AI is producing part of a presentation script, the student says, “I need something like this. Here’s what I’m trying to accomplish.” And the AI comes up with the solution and content for it. All of that is level five
3 Considerations for Selecting a Use Level
The three things that we have to think about, however, when we’re determining an appropriate level.
Assignment Purpose
First, what’s the purpose of the assignment? What why are we giving the assignment to the students?
Cognitive Demand
The second is what level of mental processing does it require? A higher level of mental processing and analysis and synthesis can be assigned to a higher level of AI use, as well, versus just recall, which is level one.
Student Readiness
And, certainly, how ready are the kids? Can they control this process? Can they collaborate? Can they delegate? Can they oversee a process and take ownership of it?
The more capable a student is of doing those things, the more capable a student is of using the higher level of AI.
Supporting Resources
So I’ve got a couple of resources for you.
The one which directly relates to this is a guide on implementing the five levels. A companion piece to that is strategies and an approach for making sure that students aren’t cheating by using AI.
And you can get both of those over there on our website.
Conclusion
So there you go. The five levels of AI use in our classrooms by students.
I hope you found this useful. Take care.