2 Avoiding Pl(AI)giarism: How ChatGPT can be Utilized by Students and Assessed by Teachers
Tara Carey
Abstract
Students have access and are likely to use ChatGPT for educational purposes. This chapter will explore ways students can do this to enhance their learning and how teachers can rethink their assessment practices to incorporate this new technology. Instead of being held back by the fear of negative uses of AI tools, like ChatGPT, you will be encouraged to explore the many benefits it can bring to learning. ChatGPT, specifically, is versatile, user-friendly, and can perform numerous different tasks, making it ideal for student use (Estrellado & Millar, 2023).
Figure 1
Note. Tara Carey (2024) generated this image using the Midjourney AI platform. I dedicate any rights I holds to this image to the public domain via CC0.
Introduction
The widespread use of the internet and access to cell phones has created a shift in teaching and assessment. Content delivery is now just one small aspect of learning (Naidu, 2024). Traditional assessments often included demonstrated memorization of facts (Larson & Lockee, 2014). However, there has already been a significant shift towards more open-ended, alternative assessments. The primary concern with learners having access to AI tools, like ChatGPT, is the ability to cheat or plagiarize, and therefore, become apathetic in their learning (Avramovic & Avramovic, 2024). This chapter aims to explain how ChatGPT can assist in learning and create more equitable learning opportunities when used appropriately. The purpose of education is to prepare our students for the future. As educators, we play a crucial role in this preparation, and, like it or not, AI is being used in many industries worldwide. We are doing a disservice to our students by not teaching them the value and proper applications of this new technology (Estrellado & Millar, 2023). Furthermore, AI can remove barriers to learning, like limited access to teachers or tutors, language issues, or missing prerequisite skills, like writing or knowledge of a specific topic (Naidu, 2024) (Avramovic & Avramovic, 2024).
ChatGPT is a tool, not unlike a calculator, that can and will be used regardless of an instructor’s wishes. ChatGPT attempts to mimic human language while performing complex tasks based on user prompts. It is useful to students because even its free version is powerful and can perform a variety of functions, from having a conversation or brainstorming ideas to writing an essay or creating an image. It is user-friendly and prompts can be edited within the chat. It is adaptable and accessible, making it a top choice for anyone new to AI (Estrellado & Millar, 2023). Other AI tools, like Google Gemini and Microsoft Copilot, can perform similarly to ChatGPT but are currently not as widely known. However, most of the information presented in this chapter can be applied to any large language model (LLM).
Learning Objectives
Learning Objectives
-
Understand how students can make use of ChatGPT
-
Recognize potential limitations and risks of using ChatGPT
-
Understand how ChatGPT can be leveraged to alleviate pressure on teachers to provide one-on-one instruction
-
Contemplate the assessment of work that is created or assisted by ChatGPT
Case Study
Utilizing ChatGPT for Learning
Because ChatGPT is incredibly easy to use, there are several ways students can use the application during the “assessment as learning” phase. Some of these include:
- Proofreading and providing feedback
- Creating summaries of a text
- Curating a resource list at a specific level
- Creating self-evaluation rubrics
Proofreading and providing feedback allows students to receive immediate, focused, feedback that can be customized to a student’s needs. Teachers do not always have time to proofread drafts and cannot return feedback as quickly. Peer feedback is not always helpful, so ChatGPT can create equity for students who do not have someone to proofread for them. However, students may need to be coached on how to create effective prompts to avoid ChatGPT rewriting large amounts of their work. ChatGPT can be a great resource but has the potential to encourage plagiarism (Estrellado & Millar, 2023).
Prompt Examples
“Provide 3 pieces of general feedback for the following written response:”
“List 3 ways I can improve the writing quality of the following text”
“Proofread for areas that lack clarity”
These examples will prompt ChatGPT to evaluate things like clarity, sentence structure, and real-world examples.
Non-Example:
“How can I make the following text better?”
This can result in ChatGPT rewriting the text instead of providing feedback.
Creating summaries of a text can help students gain a deeper understanding of information they did not fully understand and can be a great supplement to teacher instruction (Avramovic & Avramovic, 2024). Students at different reading levels can also self-differentiate by having information paraphrased at any level they choose. Since assessment is about what has been learned (Larson & Lockee, 2014), finding different ways to learn the material should be encouraged and applauded.
Curating a resource list at a specific level can significantly reduce the time and frustration a student spends on an assignment. Just like finding information in an encyclopedia is outdated, maybe so too are the skills needed to find information with a Google search. ChatGPT can not only create a list of resources but can also give a description and reason why that resource was chosen. Students can then decide which resources they want to explore further. One thing that ChatGPT does not do well is make sure the links work. In a generated list of 10 resources for middle years genetics, only three links directed me to another page and one was a no longer available video. The others could be copied and pasted into a search bar with minimal extra effort. An example is shown below. The full list can be viewed here.
Resource List Example
-
Learn Genetics by the University of Utah
-
Website: Learn Genetics
-
Description: This site provides easy-to-understand animations, videos, and interactive tutorials on key genetics topics like DNA, genes, heredity, and traits.
-
Why It’s Good: Clear explanations, interactive features, and accurate scientific content.
-
Genetics and Heredity (Khan Academy)
-
Website: Khan Academy Genetics
-
Description: Khan Academy offers free instructional videos on heredity and genetics that are concise and easy to follow.
-
Why It’s Good: Short, engaging videos with quizzes and exercises to reinforce learning.
Note. Tara Carey (2024) generated this list using the ChatGPT platform. I dedicate any rights I hold to this image to the public domain via CC0.
Creating self-evaluation rubrics for open-ended or individualized assignments is something that many students struggle with. ChatGPT will not make a good rubric without a well-thought-out prompt so this encourages students to think about the main points of their assignment without getting bogged down by the specific wording. These rubrics can be used as formative assessments, whose purpose is to show gaps in knowledge and skills (Larson & Lockee, 2014) or summative assessments to show mastery of an outcome.
Still not convinced that ChatGPT is useful in the classroom? Think about the following scenario:
Vignette
Two Grade 9 students are completing a research assignment on the development and application of CRISPR, a relatively new technology that allows for the elective modification of DNA in living organisms. The learning outcome is to understand how the technology works and discuss it’s implications on society.
Student A types “CRISPR” into the Google search engine and hundreds of results show up. She chooses the top result and finds that the information is at a very high level and she can’t understand most of it. She spends the next 20 minutes clicking on websites looking for one that is more suitable. The site has tons of information but most of it is not what she needs for her assignment. Her finished paper is full of technical terms she doesn’t fully understand, spelling and grammatical errors because her parents were unable to proofread her work before the due date, and only touches on the possible impacts on society.
Student B uses ChatGPT to generate a resource list, using a prompt that requests information specific to the role of CRISPR in the future. A list of 10 resources is created, and he looks through the brief descriptions to choose which ones to look at first. One site seems to have a lot of useful information but there is a significant amount he doesn’t fully understand so he asks ChatGPT to summarize the information at a middle school reading level. Now Student B can confidently write his paper. Once finished, he asks ChatGPT to provide 2 pieces of feedback to improve his paper. The AI highlights some spelling errors and tells him he should include more specific examples on the implications for society, which he then adds. He also asks to be graded on the teacher-given rubric. ChatGPT indicates that points would likely be lost on an incomplete explanation of how CRISPR works. Student B does more research and adds the information before turning in the completed assignment.
Could the learning outcome be met by both scenarios? Who do you think learned more? Did AI compromise the integrity of the assignment?
Assessing ChatGPT-Assisted Assignments
Assessment techniques have been shifting toward more alternative, individualized and adaptive methods for years, emphasizing aligning these assessments with learning outcomes (Larson & Lockee, 2014). Focusing on evaluating critical thinking and problem-solving skills is where educational research has already taken the field of education (Estrellado & Millar, 2023), so it isn’t much further of a step to look at how these same skills can be assessed while using ChatGPT or other AI programs. The only question an instructor needs to ask themselves is, “can a student show mastery of this outcome while using AI?” The answer will help determine to what extent ChatGPT can be utilized in a classroom. Instructors also need to be clear on the expectation of their students to acknowledge where and how AI was used during an assessment.
One method to encourage proper use of ChatGPT is to assess the entire process, not just a final product. For written work, have students enter their first draft into ChatGPT with the prompt of asking for feedback. The first draft and AI-created feedback can be submitted with the final assignment. You could even take it further by having students choose one or two areas for specific feedback. This way, you can assess their choices and where they thought they needed improvement. For research, students can ask ChatGPT for a list of 10 resources, then choose the most useful ones. When creating artwork, assess the creativity of the prompt they used, or have them submit multiple image generation attempts and explain why they chose the final one. Table 1 gives an example of a rubric that incorporates the use of AI in a research project. The rubric can also be found here.
Table 1
Rubric Incorporating AI Use
| Criteria | Exemplary | Proficient | Developing | Needs Improvement |
| AI Brainstorming Usage | Effectively uses AI for brainstorming, with clear documentation of AI-generated ideas. Demonstrates critical thinking in evaluating and selecting relevant AI-generated ideas to move forward with. | Uses AI for brainstorming and documents the process. Demonstrates some critical evaluation of AI-generated ideas but lacks depth in idea selection. | Uses AI for brainstorming but provides minimal documentation. Limited critical thinking in the evaluation of AI-generated ideas. | Uses AI superficially or does not use AI as directed. Little or no documentation of AI usage or evaluation process. |
| Idea Selection & Justification | Provides a clear, well-reasoned justification for chosen idea(s) based on personal insight, research, and AI suggestions. Demonstrates independent decision-making. | Provides a justification for the chosen idea(s), showing some connection between personal insight, research, and AI suggestions. | Provides a basic explanation for idea selection but lacks clear connection between personal decisions, research, and AI input. | Little or no justification for chosen idea(s). Does not demonstrate independent decision-making. |
| Research Depth | Thorough, well-researched exploration of the chosen idea, incorporating diverse sources. Demonstrates depth of analysis and understanding of the subject matter. | Adequate research, using relevant sources, but some areas lack depth or variety. Analysis is sound but not as comprehensive. | Minimal research, with limited sources or a narrow exploration of the chosen idea. Analysis lacks depth. | Insufficient research or lacks variety of sources. Analysis is shallow or incomplete. |
| Creativity and Originality | Demonstrates high levels of creativity and originality in developing the idea. Takes unique or unexpected approaches in advancing the project beyond AI suggestions. | Demonstrates creativity and originality in developing the idea but relies somewhat on AI suggestions without significant extension or adaptation. | Shows some creativity, but the idea is largely derivative of AI suggestions without significant original contribution. | Lacks creativity or originality in idea development. Heavily reliant on AI with minimal personal input. |
| Critical Reflection | Provides thoughtful and reflective critique of the AI’s role in the brainstorming process. Discusses strengths and limitations of AI and how it influenced decision-making. | Provides reflection on the AI’s role, noting its strengths and limitations. Reflection is somewhat shallow but acknowledges AI’s influence on decisions. | Provides limited or vague reflection on AI’s role, with minimal insight into how AI influenced the decision-making process. | Little or no reflection on the AI’s role in the project. No discussion of how it influenced the process or decision-making. |
| Presentation & Clarity | Well-organized and clearly presented project. Ideas are articulated logically, and all sections are cohesive. Uses clear language and correct grammar. | Organized and clearly presented with minor issues in flow or language. The structure is coherent but could be more polished. | Project is somewhat disorganized, with unclear transitions or sections. Ideas may lack logical flow. Some grammar or clarity issues. | Project is disorganized or unclear. Significant issues with clarity, flow, and grammar detract from overall understanding. |
Note. Tara Carey (2024) generated this table using the ChatGPT platform. CC dedicates any rights it holds to this image to the public domain via CC0.
Instructors need to be aware of the capabilities and limitations of the tool through continual professional development (Estrellado & Millar, 2023). As new updates or technologies are introduced, assessments may need to be re-evaluated to determine their effectiveness. For example, the free version of ChatGPT is unable to interpret images, diagrams, or large data files. Questions that utilize these things, like explaining a political cartoon, cannot be accomplished by ChatGPT. Similarly, it cannot currently create complex diagrams, like concept maps and is less effective at tasks that require higher levels of thinking. If this changes in the future, an instructor may want to consider changing their assessment. Avramovic and Avramovic (2024) give another suggestion of asking questions that are specific to material covered in class or on recent current events. Without relevant, updated information, ChatGPT cannot be used to replace learning.
Work that has been entirely created using ChatGPT is often easy to detect due to the general, unoriginal, responses it generates (Estrellado & Millar, 2023). Additionally, if multiple students are entering similar prompts, they will likely generate similar answers (Avramovic & Avramovic, 2024). There are also AI-detector tools that can be used with varying success. During one study, five of the ten detection tools tested could accurately identify AI-generated content, and two could detect AI-generated content even after being paraphrased by another AI program. The two most accurate tools were Sapling and Undetectable AI. The study went on to explain that these results were inconsistent with other studies conducted and that the accuracy of these tools is variable (Kar, Bansal, Modi, & Singh, 2024).
Responsible use of AI
Even when individuals are attempting to use ChatGPT as a learning tool, there are still some flaws we all need to be aware of. Chatbots, like ChatGPT, can give inaccurate or misleading information (Avramovic & Avramovic, 2024), can reinforce bias based on their training data (Estrellado & Millar, 2023) and “hallucinate” to make up data that is presented as factual. Teachers are professionals and can more easily be critical of information presented by the AI. Students need to be taught these skills and be continually reminded not to unquestioningly trust the information.
Privacy is also a concern regarding ChatGPT. It is the role of schools and teachers to help protect student privacy (Estrellado & Millar, 2023). When students are using AI, there is a higher potential for private information to be shared. However, this is not a new issue, as social media platforms and online engagement have made digital literacy an ever-growing necessity. However, teachers must be aware that uploading a student’s work onto ChatGPT without permission is unethical. The best path forward is to teach students how to use this tool effectively, and help them to understand the possible privacy issues. They can then make an informed choice whether to use AI.
Besides concerns about the information ChatGPT utilizes, there are also environmental concerns with using AI. These models use a significant amount of water during their training and daily operations, with ChatGPT using about 500 mL of water for about every 50 questions asked. Water is needed in the cooling systems in the data centers, and to produce the electricity needed to run the models. This can have devastating effects on local ecosystems and create even more water scarcity in some regions. Students and educators need to evaluate the costs versus benefits of using ChatGPT when deciding how it can enhance learning. Table 2 gives some insight into how and when the benefits may outweigh the costs. If OpenAI, creator of ChatGPT, begins investing in more energy-efficient technologies, alternative power sources, or water recycling, much of its environmental impact could be reduced (George, George, & Martin, 2023).
Table 2
Analysis of the Use of ChatGPT
|
Novelty
|
Using ChatGPT strictly for novelty to increase engagement. Other tools can complete the same task as the AI. Novelty wears off and engagement can be created in other ways. AI use is NOT RECOMMENDED.
|
|
Efficiency
|
ChatGPT can speed up the process, but the same task could be accomplished by one or more tools at a slower pace. If time allows, students may benefit from learning skills that would be unnecessary while using ChatGPT. AI use should be CONSIDERED, while remembering privacy and environmental concerns.
|
|
Tutoring
|
Enabling students to use ChatGPT in place of a teacher. ChatGPT is not human and cannot replace a qualified teacher. However it can benefit student learning when a teacher is unavailable. AI is RECOMMENDED, but attention should be given to the information it provides.
|
|
Creativity and
|
Using ChatGPT to encourage critical thinking and creativity, while performing the task not possible without AI. This is where the true benefits to learning can be found. AI use is ENCOURAGED
|
Note. Tara Carey (2024) adapted this table from the SAMR framework . I dedicate any rights I hold to this image to the public domain via CC0.
Future Research and Innovation
There is no question that ChatGPT will continue to grow and advance. As this happens, the potential for learning and the risks of learning loss will grow with it. As improvements are made, plagiarism and false information will become harder to detect. Hopefully we will successfully foster a culture of appreciation of this tool instead of a reliance on it. No assignment will ever be completely AI proof, however, creating authentic, interesting assignments that allow for AI assistance will encourage more learning and less cheating.
It would also not be surprising to see a change to assessment practices and grading systems in education. One possible method is a badge system that shows mastery of skills instead of overall grades (Larson & Lockee, 2014). This is already being done to some extent in many schools for primary grades, where outcomes are assessed on a number or letter scale corresponding to feedback on that particular outcome. Incorporating a similar structure into high school would create a need for many more changes, including criteria for advancement and post-secondary applications. However, the discomfort of change should not prevent policymakers from adapting to new situations. In many ways, the old grading system is outdated and no longer an accurate representation of student learning, if it ever was.
Summary
Note. Tara Carey (2024) created this video with HeyGen video generator.
Acknowledgements
I would like to thank the ETAD 873 class for their valuable feedback and inspiration on this chapter. I would also like to acknowledge the use of AI in the following ways:
-
ChatGPT was used for testing ideas, generating lists, and creating a rubric for this chapter, as well as assisted in making the script for the summary video.
-
Midjourney AI was used to generate the image in the introduction.
-
HeyGen video generator was used to create the summary video.
Open Researcher and Contributor ID (ORCID)
Tara Carey. https://orcid.org/0009-0004-3671-9348
Tara Carey holds a Bachelor of Education from the University of Regina and is currently completing her Master of Education in Educational Technology and Design. She has 13 experience teaching high school math and science.
References
Avramovic, S., & Avramovic, I. (2024). Exploring the Potential Benefits and Limitations of Using an AI Text- Generation Tool in Education: An Examination of ChatGPT’s Performance on Assessments. The Journal of Health Administration Education, 40(2), 193–204. https://www.ingentaconnect.com/content/aupha/jhae/2024/00000040/00000002/art00004
Estrellado, C. J., & Millar, G. (2023). CHATGPT: Towards Educational Technology Micro-Level Framework. International Journal of Science, Technology, Engineering and Mathematics, 3(4), 101–127. https://doi.org/10.53378/353035
George, A. S., George, A. H., & Martin, A. G. (2023). The environmental impact of AI: a case study of water consumption by chat GPT. Partners Universal International Innovation Journal, 1(2), 97-104. https://doi.org/10.5281/zenodo.7855594
Kar, S. K., Bansal, T., Modi, S., & Singh, A. (2024). How sensitive are the free AI-detector tools in detecting AI-generated texts? A comparison of popular AI-Detector tools. Indian Journal of Psychological Medicine, 1–4. https://doi.org/10.1177/02537176241247934
Larson, M. B., & Lockee, B. B. (2014). Streamlined ID – A practical guide to instructional design. Routledge. https://doi.org/10.4324/9780203118757
Naidu, S. (2023). Reimagining education systems: How research on digital learning can inform pedagogical practice. Communications in Computer and Information Science, 3–12. https://doi.org/10.1007/978-981-99-8255-4_1
OpenAI. (2023). ChatGPT (September 2024 version) [Large language model]. https://chat.openai.com/chat
