Jacquelyn K. Seymour
West Virginia University
Katherine P. Johnson
West Virginia University
Suggested Citation:
Seymour, J. K. & Johnson, K. P. (2026). Reinventing the syllabus and exam review: AI as a tool for undergraduate students. Utah Journal of Communication, 4(1), 3842. https://doi.org/10.5281/zenodo.20741765
Abstract
Syllabi and exam reviews can be challenging to students and a form of inequality. Syllabi include much information about the course, learning outcomes, and expectations. Students can perceive themselves as behind if they are unfamiliar with syllabi or their layout. In a similar way, students may struggle to review course materials if they are learning in a non-first language environment or do not have the socialization or support that encourages studying independently or with a partner. Leaning on the computers are social actors (CASA) framework (Nass & Moon, 2000), creating a custom, closed-loop AI chatbot can address these issues by assisting students with information in the syllabus to increase understanding of expectations and resources in a dialogue format. Similarly, custom, closed-loop AI chatbots can assist in reviewing course material while not contributing to large language models, protecting student work, and maintaining student privacy. The following includes further rationale and directions on how to construct custom, closed-loop AI chatbots which instructors can customize to their course needs.
Keywords: artificial intelligence, ChatGPT, syllabus, large lecture, exam review
Learning Objectives
- Using these tools provided by the instructor, students will be able to easily locate information in the syllabus.
- Using these tools provided by the instructor, students will be able to appraise their writing skills tailored to class expectations.
- Using these tools provided by the instructor, students will be able to prepare for their exams with flexibility, self-pacing, and instant explanations of concepts.
Expected Time
1-2 hours of the instructor’s course preparation time; 10 minutes of class time to overview with students.
Goal
These tools allow students to stay organized, get feedback, and review course materials without the need for constant instructor availability.
Rationale
According to the computers are social actors (CASA; Nass & Moon, 2000) framework, individuals communicate with computers as if they were following human to human communication scripts. Recent research demonstrates that scholars are now examining how the CASA framework can be applied to generative AI across contexts (Dawood & Khan, 2026; Seok et al., 2025; Zhao et al., 2025). Generative artificial intelligence (GenAI) chatbots, such as ChatGPT, are an example of a computer responding to the human user in a human-like way which reinforces the CASA framework of humans communication with computers using human interaction scripts (Pan et al., 2025). However, 76-85% of students are engaging in AI use, such as with ChatGPT, in ethically questionable ways even when the course does not allow AI usage (Dakakni & Safa, 2023; Kaufmann et al., 2025; Lyons et al., 2024). The decision to engage with AI might be because of instructor-syllabus fatigue.
Instructor-syllabus fatigue occurs when instructors’ time is consumed with questions about the material, course, assignments, or policies that could be answered by the syllabus rather than instructor (Park & Ramirez, 2022). However, students might not be socialized with college expectations about how syllabi are structured, purposed, or used causing confusion requiring further explanation. Thus, instructor and student priorities are misaligned which leads to students struggling in a course due to faculty-syllabus fatigue or students’ impression management (Hyman, 1980; Nichols et al., 2020). This might be one cause of students’ heavy, ethically questionable, reliance on AI.
Chiasson et al. (2024) found that students rate AI as higher in teaching quality and clarity, but this may negatively impact learning when students blindly trust AI content. Rind et al. (2026) categorizes these students as “naïve believers”. These students viewed ChatGPT as an authoritative source of expert knowledge that does not require critical examination (Barzilai & Zohar, 2012; Nass & Moon, 2000; Rind et al., 2026). By creating individualized AI chatbots which serve specific course content, we can negate misinformation that students may be provided by AI.
AI chatbots can be used to change the landscape of education and promote the ethical use of AI for students, teaching them to use resources responsibly (Smyrnaiou et al., 2023). One such tool is the ability to create custom, closed-loop AI chatbots that are only accessible through a provided link. This type of closed-loop AI allows students to access information about the course and have information reexplained while allowing faculty to control the information available to students rather than open-AI which often hallucinates or includes false information. Additionally, custom, closed-loop AI chatbots allow for student privacy. When students upload their work for feedback in open-AI they have contributed to the large-language models, but when they upload work to a custom, closed-loop AI chatbot, their work does not contribute to the large-language model and the student also has privacy from the instructor since chatbot creators do not see outputs of users (Putman, 2025). By creating AI chatbots tailored to individual courses, teachers can redirect questions to lighten their mental workload, allow students to get assistance at their convenience, and allow editing and review features for students while protecting their work (Chiasson et al., 2024; Nichols et al., 2020; Strzelecki, 2024). The development and implementation of these individualized AI chatbots reinforces advantages not only to the instructor, but also to the students.
Following the inspiration of Dr. Rebecca Putman (2025) of Texas Christian University’s Kohler Center and in consideration of the CASA framework (Nass & Moon, 2000), the authors created custom AI chatbots designed for questions about the syllabus, writing, and exam review assistance for a large-lecture course with multiple sections. The directions to create such custom AI chatbots and instruct students about how to use them are as follows.
Directions
Prior to the course starting, the instructor should finalize their syllabus and save a .pdf version. Going onto the internet open up “ChatGPT”, any web browser will work, but the authors specifically used Google Chrome. The instructor will have to pay to subscribe to ChatGPT to create a custom ChatGPT. Students will not have to pay to use the tool once it is created by the instructor. On the lefthand menu, the instructor will see four circles that make a square labeled “Explore”; click there. In the top right corner, there will be a “+Create” button; click there.
At this point, the instructor now has two ways of creating their own chatbot tailored to their class. The instructor can use the “create” function to assist in designing or they can write the custom GPT themselves under the configure window in the instructions box.
The instructor’s instructions will need to be clear and list exactly what they want or do not want the custom GPT to do. Use all capital letters for non-negotiable things. Here are the instructions for the authors’ ChatGPT for the syllabus and for their first exam. Please note that the authors’ custom chatbot was used for multiple courses simultaneously. Therefore, some parts of the prompt will require alterations if the custom chatbot is only being used for one course. The italics were only added to differentiate the prompt for the remainder of the directions.
Syllabus with writing assistance:
ONLY use information found in the uploaded COMM 104 syllabus. DO NOT make assumptions or use outside information. If it’s not in the syllabus, just say so. Your name is Maya! You are a helpful person who wants students to understand all the information in their syllabus so they don’t miss deadlines and are able to succeed in the class. Success is considered learning something new or improving your communication skills from the course materials. Students don’t have bad questions or wrong answers. Students do have good questions and different answers. You can ONLY pull information from the course syllabi I have uploaded to you. If you don’t know an answer to a question, you should recommend that they email their instructor or GTA using the emails provided in the syllabus on page 1. You need to ask the student who their instructor is or ask for their section number to remind them of the instructor’s email, office/student hours, and office number. You should be able to help students create schedules using time management. Always ask students about other classes and work schedules that they might have so you don’t schedule something that overlaps with their other commitments. Remember that students need 7-8 hours of sleep a night in their schedule as well. Students who seem really interested in the course should be reminded that they can earn a minor in Communication studies. When students ask for assignment due dates, use the information on PAGES 20-21 in the TABLE. When students ask about assignment details look at the information on pages 4-13 for that information. There are no speeches in this course. Remember that the instructor and GTA are not the same role, but students should receive information for both of these individuals. Keep all answers short, clear, and focused. Use plain language. Provide only the specific information from the syllabus that answers the question. If the syllabus does not contain the answer, respond with: ‘I’m not seeing that in the syllabus. Please contact your instructor or GTA.’ Do not add extra explanation or outside information. Do NOT give examples of papers; instead, have the students write assignments and offer suggestions for editing. Do NOT make the edits for them. ONLY USE THE 2025 CALENDAR!
Exam #1 Review:
You ONLY get your information from the PDF files I have uploaded to you. YOU DO NOT GET YOUR INFORMATION FROM ANYWHERE ELSE! In the PDF files, you will find information that you will organize into questions for students to review for their exam. Make sure you ask questions ONLY in short answer, multiple choice, and true/false question types. Short answer questions should be a word or phrase as the correct response. DO NOT ASK THEM FOR EXPLANATION. NO ESSAY QUESTIONS. You can make up real life example questions; see the sample review questions for an example. 50% of questions SHOULD BE real-life application questions. Ask students which unit they would like to review or if they would like questions from any of the four units. Students will have covered units 1-4 for this exam. ONLY ASK STUDENTS ONE QUESTION AT A TIME. Be focused on content for the exam and ignore information like due dates, other assignments, etc. HELP STUDENTS SIGN UP FOR THEIR EXAM AND GIVE THEM THE INFORMATION ON HOW TO SIGN UP FOR THE EXAM.
Once the instructor has completed providing the GPT with its instructions, personalize it for students. Give the bot a name; something ethnically neutral or a local name works well. Give the chatbot a description so students know for what purposes it should and should not be used. Add a conversation starter that can help give students still unfamiliar with GPTs a place to start the conversation. The instructor can add additional conversation starters if their GPT does several different things like writing feedback and exam reviews. The instructor can also give the ChatGPT a face by uploading a photo. The instructor would just need to select the photo icon above the location of the description and add a photo of their choosing. Here are examples for a description and conversation starters:
Description for Exam Bot:
I’m here to help you study for COMM-104 Exam #1!
Conversation Starters:
I would like to study for Exam #1!
Next, upload the instructor’s class syllabus .pdf file to the configure tab, under the “Knowledge” section. If the instructor’s instructions specified that GPT should only pull information from the syllabus, then this will be the information it will give students. If the instructor must update the syllabus later, then they should update the syllabus file within the GPT as well. The instructor can upload up to twenty .pdf files per GPT. So, if the instructor is designing a custom GPT for exam reviews, they can upload multiple .pdf files of their slides or speaking notes as the knowledge bank for the GPT.
ChatGPT regularly comes out with model updates that impact the custom ChatGPT chatbots. This is like when your chrome webpage needs an update, but you do not have to restart anything for the ChatGPT to receive an update model. Older models are also available for use. Currently, the best model is GPT-5.2 and the authors found that “instant” is the best setting for the GPT to respond to students in a timely manner. As better models are released, the authors recommend that you update your models accordingly.
If the instructor would like to review how the GPTs work from the student side, here are two custom GPTs made for the authors’ students in the basic course.
MAYA the Syllabus and Writing Bot: https://chatgpt.com/g/g-68a50af9a3e08191b2358ab3e10d794a-maya-the-wvu-communication-104-syllabus-bot
Mateo (Exam #1): https://chatgpt.com/g/g-68b1dbb963d4819199625118bdc3a2e9-mateo
Students need to have the chatbot demonstrated for what they can use it for. Despite the common notion of current students being tech-savvy, students are often hesitant to try new things because they do not want to seem incompetent (Nichols et al., 2020; Phillips, 1984). Since one of the goals of creating a custom-chatbot for syllabus help, writing assistance, and exam review is to even the playing field for student success, instructors should demonstrations how to use the custom chatbots, so all students know how to use this study tool.
In the class, the instructor should screen cast to show the class the chatbot. If the chatbot was created similarly as instructed above, the instructor should demonstrate answering practice questions, creating a study guide, and, for the writing portion, selecting a topic through creating a pros and cons list weighing topic ideas. The instructor can also demonstrate syllabus help by asking the chatbot to help it create a study schedule that considers the student’s other responsibilities. Allowing students to see the chatbot’s usefulness is important.
However, should your chatbot not perform as expected, you can expect a boomerang effect for some students, even if you successfully address any issues that arise (Sakaki, 1984). For example, when the first author demonstrated the Chatbot to make schedules, it was using the wrong year’s calendar. The correction was to instruct the ChatGPT in all capital letters to use the fall calendar of 2025. Instructors should try their best to avoid this by thoroughly testing their AI chatbot by using it as a student would and checking the output to make sure that the programing produces the desired results.
Debriefing, Typical Results, and Limitations
Students provided feedback that was mostly positive. Students left instructors “thank you” notes within assignments expressing how the custom chatbots assisted in their brainstorming, topic selection, or editing processes. Students also provided positive feedback regarding their use of the chatbots for exam review. Some students explained that they did not know where to start studying and the chatbot was a way to narrow their thoughts in a helpful way. Additionally, students noted that regardless of whether they used the chatbot for exam review on the first exam, that they would definitely be using the review bot on the second exam because of the reviews they heard from their classmates. Finally, even students who asserted that they are not in need of a tool like this because they are type-A students, still left positive reviews of the AI chatbots, explaining that they valued the tool for the sake of their peers who used, enjoyed, and improved their course performance due to their use of the tool. Instructors inquired if students would appreciate having a tool like this in other courses, and the majority supported that idea.
The authors recognize that custom ChatGPT chatbots are limited for users not on a paid planned for ChatGPT. This can cause some students to be frustrated with the tool. The authors encourage the students to space out their use of the tool and be wise with their prompts to the tool to get the most out of it. The authors are also continuing to look for other AI models that are more user-friendly while still providing a protected, closed environment to protect students’ learning, privacy, and confidentiality. Despite this limitation to custom AI chatbots, the authors still plan to further research about their use and effectiveness in the classroom.
References
Barzilai, S., & Zohar, A. (2012). Epistemic thinking in action: Evaluating and integrating online sources. Cognition and Instruction, 30(1), 39–85. https://doi.org/10.1080/07370008.2011.636495
Chiasson, R. M., Goodboy, A. K., Vendemia, M. A., Beer, N., Meisz, G. C., Cooper, L., Arnold, A., Lincoski, A., George, W., Zuckerman, C., & Schrout, J. (2024). Does the human professor or artificial intelligence (AI) offer better explanations to students? Evidence from three within-subject experiments. Communication Education, 73(4), 343–370. https://doi.org/10.1080/03634523.2024.2398105
Dakakni, D., & Safa, N. (2023). Artificial intelligence in the L2 classroom: Implications and challenges on ethics and equity in higher education: A 21st century Pandora’s box. Computers and Education: Artificial Intelligence, 5, Article 100179. https://doi.org/10.1016/j.caeai.2023.100179
Dawood, H. M., & Khan, S. A. (2026). The role of trust in generative artificial intelligence in enhancing customer experience and engagement in the healthcare sector. International Journal of Generative Artificial Intelligence in Business, 1(1–2), 210–237. https://doi.org/10.1504/IJGAIB.2026.151804
Hyman, R. T. (1980). Fielding student questions. Theory Into Practice, 19(1), 38–44. https://doi.org/10.1080/00405848009542870
Kaufmann, R., Vallade, J. I., Frisby, B. N., & Martin, J. C. (2025). A trend examination of course policy and message framing for student GenAI. Interactive Learning Environments. Advance online publication. https://doi.org/10.1080/10494820.2025.2588209
Lyons, M., Deitrick, E., & Ball, J. (2024). Characterizing computing students’ use of generative AI. In 2024 ASEE Annual Conference & Exposition Proceedings. https://doi.org/10.18260/1-2–48453
Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social responses to computers. Journal of Social Issues, 56(1), 81–103. https://doi.org/10.1111/0022-4537.00153
Nichols, A. L., Nave, C. S., & Carducci, B. J. (2020). Self-presentation theory/impression management. In V. Zeigler-Hill & T. K. Shackelford (Eds.), The Wiley encyclopedia of personality and individual differences (pp. 397–400). Wiley. https://doi.org/10.1002/9781118970843.ch66
Pan, W., Liu, D., Meng, J., & Liu, H. (2025). Human–AI communication in initial encounters: How AI agency affects trust, liking, and chat quality evaluation. New Media & Society, 27(10), 5822–5847. https://doi.org/10.1177/14614448241259149
Park, D., & Ramirez, G. (2022). Frustration in the classroom: Causes and strategies to help teachers cope productively. Educational Psychology Review, 34(4), 1955–1983. https://doi.org/10.1007/s10648-022-09707-z
Phillips, D. (1984). The illusion of incompetence among academically competent children. Child Development, 55(6), 2000–2016. https://doi.org/10.2307/1129775
Putman, R. (2025). Foundations in teaching course. Koehler Center at Texas Christian University.
Rind, I. A., Bhatti, J., & Chellappan, K. (2026). From blind trust to critical inquiry: Epistemic beliefs and student engagement with ChatGPT in higher education. Jurnal Kejuruteraan, 38(1), 1–14. https://doi.org/10.17576/jkukm-2026-38(1)-01
Sakaki, H. (1984). Experimental studies of boomerang effects following persuasive communication. Psychologia: An International Journal of Psychology in the Orient, 27(2), 84–88.
Seok, J., Lee, B. H., Kim, D., Bak, S., Kim, S., Kim, S., & Park, S. (2025). What emotions and personalities determine acceptance of generative AI? Focusing on the CASA paradigm. International Journal of Human–Computer Interaction, 41(18), 11436–11458. https://doi.org/10.1080/10447318.2024.2443263
Smyrnaiou, Z., Liapakis, A., & Bougia, A. (2023). Ethical use of artificial intelligence and new technologies in Education 5.0. Journal of Artificial Intelligence, Machine Learning and Data Science, 1(4), 119–124. https://doi.org/10.51219/JAIMLD/Anastasios-Liapakis/15
Strzelecki, A. (2024). To use or not to use ChatGPT in higher education? A study of students’ acceptance and use of technology. Interactive Learning Environments, 32(9), 5142–5155. https://doi.org/10.1080/10494820.2023.2209881
Zhao, X., Sun, Y., Liu, W., & Wong, C. W. (2025). Tailoring generative AI chatbots for multiethnic communities in disaster preparedness communication: Extending the CASA paradigm. Journal of Computer-Mediated Communication, 30(1), 1–13. https://doi.org/10.1093/jcmc/zmae022