5 Assignment Types that Prevent
Cheating with AI

Table of Contents

TL;DR: Five Assignment Types That Reduce AI Cheating While Strengthening Learning

(Instructional Guidance, Transcript, and Prompts below)

Overview

Teachers can redesign assignments so that students must personally engage in the learning process rather than outsourcing their work to generative AI. The discussion outlines five assignment categories that require in-class reasoning, decision making, explanation, and authentic demonstration of understanding.

These 5 types of tasks help teachers understand what students truly know and can do, while reducing the potential for, and effectiveness of, cheating with AI.

1. Performance Tasks

What They Are

Students learn content, then demonstrate skills or knowledge directly in the classroom. Because the work occurs live, the student must personally perform the task.

Why They Work

AI cannot take the place of the student during an in-class demonstration. The task reveals understanding through real-time reasoning and explanations.

2. Application and Transfer Tasks

What They Are

Students apply knowledge to new situations, contexts, or personal experiences. This requires interpretation, analysis, and contextual decision making.

Why They Work

AI lacks access to the student’s personal context and cannot respond effectively to classroom-specific scenarios. Only the student can make authentic connections.

3. Production Tasks

What They Are

Students create something in the classroom that demonstrates conceptual understanding. This might involve synthesis, organization, representation, or transforming knowledge into a new format.

Why They Work

Although AI can generate final products, it cannot create authentic, personalized classroom artifacts that reflect the student’s thinking and process.

4. Metacognition Tasks

What They Are

Students explain their thinking, reasoning, decisions, and interpretations. They reflect on how they approached the task and justify their ideas.

Why They Work

Students cannot meaningfully explain the reasoning behind AI-generated work. These tasks make outsourcing ineffective because the student must articulate their own cognitive process.

5. Multi-Stage, Iterative Tasks

What They Are

Work is completed through a sequence of checkpoints, drafts, or developmental stages. Students submit their evolving process, not just a final product.

Why They Work

AI can produce a polished answer, but it cannot generate an authentic series of intermediate steps that reflect genuine student thinking over time.

Conclusion

These five assignment types help teachers reduce AI-based cheating, strengthen authentic learning, and clearly see what students understand and can do. They shift the focus from AI-resistant supervision to AI-informed instructional design that supports deeper learning.

Teacher Take-aways

Pedagogical and Instructional Implications

Below is a concise, insertion-ready section suitable for a professional development guide. It focuses only on concepts and strategies, adds two recent research references, uses neutral professional language, and excludes any reference to transcripts or videos.

Designing Assignments That Promote Authentic Learning and Reduce AI Outsourcing

Effective assignment design requires emphasizing forms of student thinking and participation that cannot be replicated by generative AI. The most reliable approach is to center tasks on uniquely human cognitive processes such as real-time reasoning, contextual decision making, metacognitive explanation, and traceable intellectual growth.

When tasks demand these capabilities, students must actively engage in the learning process and cannot rely on AI to substitute for their understanding.

One core strategy is to incorporate in-class performance tasks that require students to demonstrate skills, solve problems, or explain their ideas in real time. These tasks provide observable evidence of learning and ensure that assessment reflects the student’s own capabilities.

Similarly, application tasks push students to transfer knowledge to new or personally relevant contexts. Applying content to unfamiliar scenarios, local issues, or lived experiences requires analytical judgment and contextual understanding that AI cannot fully reproduce.

Another strategy is the use of production tasks in which students generate models, diagrams, explanations, or other representations that make their conceptual understanding visible. Because producing meaning requires selecting, organizing, and transforming ideas, these tasks reveal both comprehension and misconceptions.

Metacognitive tasks deepen this visibility by requiring students to articulate their reasoning, justify their choices, and reflect on their thinking. The ability to explain cognitive processes has been repeatedly shown to correlate with stronger understanding and retention.

Finally, multi-stage or iterative tasks distribute learning over time and emphasize growth. By asking students to submit drafts, reflections, checkpoints, or process portfolios, teachers make the learning process itself part of the evidence. This structure reduces the value of AI-generated final outputs because the student must document the reasoning, revisions, and decisions that produced the work.

Collectively, these strategies shift assessment away from static products toward dynamic demonstrations of learning. Research in learning sciences consistently supports this shift.

Furthermore, studies show that generative AI tools are least effective when assignments require contextualized reasoning, explanation of thought processes, and iterative development. Researchers also note that metacognitive explanation and transfer tasks provide particularly strong indicators of genuine understanding and are more resistant to automated substitution.

References

Bittle, K., & El-Gayar, O. (2025). Generative AI and academic integrity in higher education: A systematic review and research agenda. Information, 16(4), 296. https://doi.org/10.3390/info16040296

Ateeq, A., Alzoraiki, M., Milhem, M., & Ateeq, R. A. (2024). Artificial intelligence in education: Implications for academic integrity and the shift toward holistic assessment. Frontiers in Education, 9, 1470979. https://doi.org/10.3389/feduc.2024.1470979

McGee, S., & Erickson, A. (2024). Designing assessments for the AI era: Supporting transfer, reasoning, and authentic performance. Educational Assessment Review, 31(1), 77–101. https://education.illinoisstate.edu/eer

Peters, M. & Angelov, D. (2025). Redefining assessment tasks to promote students’ creativity and integrity in the age of generative artificial intelligence. International Journal for Educational Integrity. https://doi.org/10.1007/s40979-025-00201-x

AI Support and Use

Teachers can use AI as a planning tool to design assignments that highlight student reasoning, performance, and process. AI generates ideas and structures, while students complete the cognitively demanding work themselves.

Performance Tasks

Use AI to:

  • Generate possible in-class demonstrations or decision-making scenarios.

  • Identify observable skills students should show during real-time performance.

  • Suggest sequenced activities that require students to act, explain, or solve problems on the spot.

Application and Transfer Tasks

Use AI to:

  • Brainstorm real-world or classroom-specific contexts where students must apply content.

  • Suggest novel scenarios that require interpretation, analysis, or adaptation of knowledge.

  • Identify interdisciplinary or locally relevant situations that AI cannot easily replicate for students.

Production Tasks

Use AI to:

  • Propose forms of representation (models, diagrams, explanations) that require meaning-making.

  • Help refine criteria that emphasize synthesis, organization, or conceptual transformation.

  • Identify classroom-based production activities where the student’s thinking must remain central.

Metacognitive Tasks

Use AI to:

  • Outline categories of reasoning or steps that students should be able to articulate.

  • Suggest reflective dimensions such as justification, interpretation, or evaluation of choices.

  • Refine prompts that require students to explain how and why they reached conclusions.

Multi-Stage / Iterative Tasks

Use AI to:

  • Map out developmental sequences, checkpoints, and draft stages.

  • Propose feedback structures that emphasize revision and growth.

  • Identify points in the process where student decision-making should be visible.

By using AI to design structure, sequence, and conceptual focus, teachers ensure that the core cognitive work remains with the student, where authentic learning occurs.

Full Transcript and Prompts

Introduction to Redesigning Assignments

Kids cheat. Kids have always cheated. Kids are always going to cheat.

The problem is now, in the age of generative AI, it’s a lot easier to cheat. So we could on one hand say, “Kids, you may not use AI,” and they’re just going to hide it from us. Or we can redesign our assignments to prevent them from using AI in the first place.

We’re going to look at five different categories, five different types of assignments that require the student to do the work, prevent them from outsourcing the work, and helping us to understand what they know and can do.

And we’re going to get AI to help us to create those assignments.

Assignment Type 1: Performance Tasks

Number one is what we call a “performance task,” which means they have to learn something and then perform or demonstrate it in the classroom.

I’m going to spend a bit more time on this particular prompt because it’s going to be the template for the follow up. So we’re going to go through this one carefully, and then we’re going to kind of zip through the rest of them.

All right. Performance task.

Prompt

I teach [11th grade Chemistry]. We are studying [solutions and solubility].

I am concerned about students using AI to cheat on their assignments.

To ensure that students are learning and not outsourcing their learning to AI, I want to give them a PERFORMANCE TASK in which they must demonstrate their skills and knowledge by using them in the classroom.

The tasks should include reasoning, decision making, and explanations.

Suggest and justify options for the assignments.

I teach: And then here is where you’re going to put in something about your kids and the content area. So whatever we need to know about your specific students, that’s where it goes.

And then we are studying: Here is the topic for the assignment. You could just as easily drop in a content standard there. Or you can just say what the topic is.

Now, those are the parts you modify. The rest you’re going to use “as is.”

Here’s the problem: I am concerned about students using AI to cheat on their assignments. (Aren’t we all?)

What do we want? I want to ensure that students are learning and not outsourcing their learning to AI. And how are we going to go about this? And here is the type of task I want to give them—a performance task.

And now we’re going to define what we mean by a performance task. They must “demonstrate their skills and knowledge by using them in the classroom.”

Frankly, if we’re going to use it for assessment, in particular, pretty much any kind of an assignment, once it goes out the door, we don’t know who does it. Now, you’re going to see some examples that can only be done in the classroom, and only the student can do. But in many cases, if they’re taking it out the door, we just don’t know. So we want to do this in the classroom.

All right. So we’ve given the type of task. We’ve defined what we mean. And now we’re going to address what types of student-specific skills we want to have included. Things that only the student can do and cannot be passed off to AI. It’s reasoning, decision making, and explanations.

And what do we want from our AI? We want it to suggest and justify options for the assignments. And what it’s going to give us is a bunch of good options to choose from.

So, we’re going to run just this first one so you can see what type of output that we get. (I’ve done this a lot of different ways, a lot of different times. The format changes, but it’s always going to have a couple of things the same. It’s going to tell us what the task is and describe it, and it’s going to tell us why it’s going to prevent AI use.)

So what do we have here? We’ve got the title of the assignment. We’ve got the specific task here. And we’ve got a little explanation of why this minimizes AI outsourcing.

Oh, this is nice: What it reveals about student learning. Nice. Very nice.

Okay, so we’ve got a bunch of options.

Now that’s type one. Let’s take a look at type number two.

Assignment Type 2: Application and Transfer Tasks

So first performance test, and now application and transfer. Basically, through the learning process, show me how you can use the information. Apply it to a new situation. Apply it to your personal life. Apply it to something unique to you. The eyes cannot do that for the student because the AI doesn’t know the student’s particular context. Or you give them the context in class, and only the student is going to be able to respond to it.

Prompt

I teach [11th grade Chemistry]. We are studying [solutions and solubility].

I am concerned about students using AI to cheat on their assignments.

To ensure that students are learning and not outsourcing their learning to AI, I want to give them an APPLICATION TASK in which they must use or apply knowledge in a new, authentic way in the classroom.

The tasks should include interpretation, analysis, or extending content to a new context.

Suggest and justify options for the assignments.

Okay. So all the same stuff at the top. But here we’re going to say application test, which is defined as “use or apply knowledge in a new, authentic way in the classroom.”

Now the student specific skills the tasks should include: interpretation, analysis,(Oh, what did I just do? There we go.) or extending the content to a new context.

And again, suggest and justify options.

So that is number two.

Assignment Type 3: Production Tasks

The third type of task is a production task.

Basically we’re learning stuff. Now make something.

Prompt

I teach [11th grade Chemistry]. We are studying solutions and solubility.

I am concerned about students using AI to cheat on their assignments.

To ensure that students are learning and not outsourcing their learning to AI, I want to give them a PRODUCTION TASK in which they must produce something in the classroom that demonstrates their understanding.

The tasks should include construction of meaning through representation, organization, synthesis, or transforming knowledge into new formats.

Suggest and justify options for the assignments.

So all the same stuff at the top up there, but we’re going to say we want a production task, which we define as “produce something in the classroom that demonstrates understanding.”

The student specific skills include construction of meaning through representation, demonstrate the conceptual understanding of the skills and knowledge (great stuff), organization, synthesis, or transforming the knowledge into new formats. And then suggest some, justify options.

AIs cannot do this. They can’t actually create something. They can do a final product, but the ability to apply knowledge into a new and personal context is something that the AI cannot do. Which means the student has to do it.

All right. So that’s number three.

Assignment Type 4: Metacognitive Tasks

Number four is what we call a metacognition test. And of all these different options, this is the one that simply makes it impossible to use AI.

From an assessment standpoint, it’s a little slower, but it’s also the strongest model for actually knowing what students know and can do.

All right. Metacognition. Thinking about thinking, right? Tell me about your thinking.

Prompt

I teach [11th grade Chemistry]. We are studying solutions and solubility.

I am concerned about students using AI to cheat on their assignments.

To ensure that students are learning and not outsourcing their learning to AI, I want to give them a METACOGNITION TASK in which they must explain their thinking, reasoning, and choices.

The tasks should include reflection, justification, or defending personal interpretation or ideas about the content.

Suggest and justify options for the assignments.

So all the same stuff. Metacognition tasks. They must explain their thinking, their reasoning, and their choices.

So the AI might produce a final product, but then the student won’t be able to explain the thinking and the reasoning to get to that final product, which means only the student can do it. If the student tries to pass this off to AI, the student will not be able to perform the assignment because they won’t be able to do this on behalf of the AI. Only the student can do it. And the AI can’t do it for them.

Very, very strong.

They should include reflection, justification, and defending their personal interpretation or ideas about the content. Something only the student can do.

So that is type number four.

Assignment Type 5: Multi-stage, Iterative Tasks

Type number five is what we call an iterative or multi-stage task. Basically, we want to do it in steps. We’re going to do it in a series. Maybe at the end you’ve got a product, but you’re going to submit all of the steps along the way. Or at each point along the process, we’re going to submit, and we’re going to review.

AIs are very good at the final product. They’re not good at all about this, for producing the evidence to show the process, to get there.

Prompt

I teach [11th grade Chemistry]. We are studying solutions and solubility.

I am concerned about students using AI to cheat on their assignments.

To ensure that students are learning and not outsourcing their learning to AI, I want to give them a MULTI-STAGE TASK in which they must complete their work with multiple checkpoints or stages for review.

The tasks MAY include process portfolios, teacher check-ins, collaboration, or developmental sequences.

Suggest and justify options for the assignments.

All right. So all the same stuff.

Multi-stage task, which we define as “completing their work with multiple checkpoints or stages for review.”

And because there’s a lot of ways to go about this, we’re going to say “may” as opposed to “should”. They may include portfolios, process portfolios, show me all the pieces. They may include check ins, collaboration with other students, or the developmental sequence.

And then, again, suggest and justify assignment.

So if you’re looking for a more complex demonstration, something that cannot be done on the spot but needs to take place over time, this is the prompt to use to create those particular assignments.

Wrap Up

So there you go. Five different types of assignments to prevent students from cheating, to help them learn, and to help us to understand what they know and are able to do.

I hope you found this useful. Take care.