The hot summer is over, and the start of the new semester is almost here. Professors may feel both the excitement of meeting your students and the concern about how generative AI can be positively integrated into the learning process. In the previous Teaching Tip No. 44, our university discussed how generative AI was used in classes, and it had the second-highest open rate (33%) among the Teaching Tips for the first semester of 2024. This Teaching Tip will review research papers on how to design classes using generative AI in different academic fields. Since ChatGPT was released in 2022, many papers on using generative AI in teaching have been published in 2023 and 2024. We looked at about 10 out of 97 papers found by searching "generative AI classes" in KERIS’s research service, RISS. These papers show how generative AI was used in writing, programming, and project-based learning for college students. We will summarize these findings to provide useful methods for professors to use generative AI.

1. Writing: Starting and Ending with Generative AI, Leading Conversations with Generative AI

  Generative AI tools like ChatGPT are very strong at generating text, which raises both hopes and concerns about using them in university writing classes. Writing is an area where generative AI can have a direct impact, and there are concerns that it might affect students' writing skills and the effectiveness of the classes. To address these concerns, Sangseok Park (2023) suggests focusing on “process-oriented writing education.” He believes that using personal experiences in writing is important and will make the writing process more dynamic.

A study by JOO MINJAE (2023) is an example of how generative AI can be used effectively in writing without harming the essence of writing. This study used generative AI in the "pre-writing stage" to set topics and objectives for research reports in a writing class at Seoul A University during the second semester of 2023. The study found that all but one of the 30 participants found ChatGPT useful for selecting research topics. Students felt that their perspectives were broadened and they gained more background knowledge, which expanded their thinking. Additionally, because generative AI can engage in conversation, it helped create a “question-answer-requestion” pattern, making it easier to get extra information or develop subtopics. However, the study also found that using ChatGPT took more time and effort to refine topics, and there were questions about whether ChatGPT helps improve the creativity of writing. Professors should consider these points when designing writing classes that use ChatGPT.

 Sukja Park (2024) suggests using ChatGPT as a "sentence correction" tool in a relatively safe way for student-led writing. “Sentence correction” means fixing grammatical errors, typos, or inappropriate expressions. ChatGPT can naturally correct foreign words to Korean and fill in missing parts of sentences to ensure subject-verb agreement. Students found that ChatGPT not only corrected spacing and spelling but also improved sentences with translation-like errors. Therefore, using ChatGPT for sentence correction can help students use it as a learning tool and as a way to reflect on their writing process without needing separate tutoring.

 Sang Min Choi (2023) emphasizes that in AI-assisted writing education, it is crucial to develop questioning skills. Students need to build their prompt engineering skills, and various methods and strategies should be developed and proposed for this purpose. "Prompt engineering" involves creating AI prompts to produce optimal results. Since the quality of the prompts determines the quality of the AI-generated outputs, students need to learn how to ask questions or give commands to AI when seeking help in writing or discussions. For example, teaching students to provide context for the results they want and how to collaborate through specific conversations is important. It is recommended that students first understand the problem they want to solve, assign roles to the generative AI, and then start the conversation. By understanding prompt engineering and learning how to collaborate with AI, students can complete tasks even with limited background knowledge. However, there are concerns about plagiarism and the need for fact-checking of information, so educational ethics guidelines should also be provided.

  Yoo Geonsu and RLEE SANGJAE (2024) emphasize the importance of learning how to create effective prompts and actively interact with generative AI. They stress that students should be guided on how to use prompts well and given plenty of practice. In real cases, students received high-quality responses when they provided detailed prompts that matched their needs. Additionally, when students first give their own writing and ask for responses based on that context, they get to review their work from a meta perspective. This experience is important as it helps achieve the ultimate goal of improving students' writing skills.

2. Programming: Fixing Syntax and Logical Errors with Generative AI and Completing Code

 The next two papers study how to use generative AI in programming classes and the learning experiences from these methods. The study by Myung-suk Lee (2024), titled "Software Education Class Model using Generative AI - Focusing on ChatGPT" developed a software education model using ChatGPT and applied it to a university Python programming class. The study then surveyed 28 students. In this model, the instructor used ChatGPT to evaluate student assignments and codes, and to check students' understanding of the class material. Students used ChatGPT to fix mainly syntax and logical errors during the "problem-solving" phase, and in the "application" phase, they used ChatGPT to develop project ideas, create algorithms, and complete their code.

Image Source: Myung-suk Lee. (2024). Software Education Class Model using Generative AI - Focusing on ChatGPT. Journal of Practical Engineering Education, 16(3), 275-282.

 After running the class for three months, it was confirmed that ChatGPT was a useful tool in programming classes. It supported tasks like grading assignments, helping with code writing, and solving errors. Students actively used ChatGPT for writing assignments, fixing errors, coding, and learning new knowledge, which improved their learning efficiency. Analyzing students' prompts showed that initially, students just copied and pasted the questions given by the instructor and got results. Over time, they started asking why the code produced those results, found errors in their own code more quickly, and used ChatGPT more actively.

 The second paper, by Jung-Oh Park (2024), titled "A study on the experience and utilization of generative AI-based classes - focusing on programming classes," investigated using an AI chatbot in a web programming class where engineering students developed a website. Students used the AI chatbot to ask common questions about source code, such as concepts, classification, necessity, purpose, operation process, and function comparison. They also asked detailed questions about adding comments line-by-line, analyzing execution issues, changing to other languages, and fixing syntax errors. The study found that students felt the AI chatbot was helpful for Q&A feedback and solving practice problems. After the middle of the semester, students had a more positive view of the chatbot. They improved their skills in analyzing source code, correcting grammar, and solving problems through the chatbot, and their expectations for learning increased. On the other hand, the study also showed that students were using the AI chatbot instead of traditional search engines to find information. The paper suggests that professors should clearly explain the ethical issues like "information falsification" and "data bias" and make sure that the specific methods for using AI-generated results and that individuals are responsible for the outcomes.

 Both papers on using generative AI in programming classes confirmed that generative AI can be a useful tool in education. However, they also emphasized the importance of teaching students how to write effective prompts and address ethical issues to use generative AI correctly.

3. Project-Based Learning: Using Generative AI to Reduce the Effort Required for Projects

  The next two papers report on using generative AI tools in student-led project-based learning, discussing their benefits and limitations.

   The first paper by Jee-Yeon Lee & Jae-Woong Kim (2024), titled “A Study on the Use of Image Processing Classes in Generative AI. Journal of Knowledge Information Technology and Systems,” explores the use of ChatGPT 3.5 in an image processing project using OpenCV. Students worked on a project to restore old photos using C++ and OpenCV, and ChatGPT provided detailed responses for setting up the environment and writing code, supporting the students’ progress. During the project, students needed to provide additional prompts to refine ChatGPT’s initial responses, indicating that the first answers alone were not sufficient. Through this process, students learned about function features and parameter meanings and adjusted their code. The study suggests that generative AI can effectively assist in project work and enhance learning. It also notes that both student effort and instructor guidance (such as prompt creation, review of prompt results, and code reviews) are essential for successful project outcomes.

  The second paper by Min Ji Young & Jeong Byoung Guk (2023), titled “A Study on the Effectiveness of Chat GPT and Midjourney in Design Education: - Focus On The Creation Of Story Illustration,” examines the use of two generative AI tools in design projects. The study involved second-year illustration students at K University, who were divided into an experimental group using generative AI and a control group not using AI. Over six weeks, the experimental group used AI tools in various stages of their design projects. Survey results showed that students used ChatGPT to create detailed stories and scenarios and Midjourney to develop ideas and refine their results. The experimental group’s results were rated higher by experts compared to the control group. The study highlights that while generative AI can support the design process, understanding its limitations and using it appropriately is crucial.

These studies confirm that using generative AI in project-based learning can significantly reduce the effort required for projects and enhance the effectiveness of the results. However, to use generative AI effectively in education, critical thinking by users and guidance from instructors are necessary to address limitations and errors. Developing clear guidelines for using generative AI is also important.

 This Teaching Tip has reviewed research on using generative AI in classes. Since 2023, there has been ongoing discussion and actual application of generative AI in university classes across various fields such as writing, coding, and project-based learning. The studies reviewed emphasize the importance of writing effective prompts and using generative AI ethically. They also stress that with proper guidance and critical use by instructors and students, generative AI can be a valuable tool for learning. We hope this Teaching Tip helps professors design effective classes using generative AI for their students' benefit.

References

Min Ji Young & Jeong Byoung Guk (2023). A Study on the Effectiveness of Chat GPT and Midjourney in Design Education: - Focus On The Creation Of Story Illustration. Journal of Brand Design Association of Korea, 21(3), 347-358.

Sangseok Park (2023). A Study on the Response of College Writing Education in the ChatGPT Era: Centered on the re-awareness and improvement of process-based writing. The Korean Journal of Literacy Research, 14(6), 259-292.

Sukja Park (2024). ChatGPT and College Writing: Learner-Centered AI Feedback : Focusing on sentence correction. The Journal of General Education, 27, 101-141.

Jung-Oh Park (2024). A study on the experience and utilization of generative AI-based classes - focusing on programming classes. Journal of Practical Engineering Education, 16(1), 33-39.

Yoo Geonsu & RLEE SANGJAE(2024). Exploring Curriculum for Critical Writing Utilizing Generative AI – Analyzing Materials of Undergraduate Students in General Education –. The Society for Korean Language & Literary Research, 52(2), 227-262.

Myung-suk Lee (2024). Software Education Class Model using Generative AI - Focusing on ChatGPT. Journal of Practical Engineering Education (JPEE), 16(3), 275-282.

Jee-Yeon Lee & Jae-Woong Kim (2024). A Study on the Use of Image Processing Classes in Generative AI. Journal of Knowledge Information Technology and Systems, 19(3), 525-531.

JOO MINJAE (2023). How to Teach Writing in the Age of Generative AI Reimagining Writing Instruction -Focusing on the analysis of university learners' perceptions of the utility of utilizing ChatGPT in the ‘pre-writing stage’. DONAM OHMUNHAK, 44, 71-103.

Sang Min Choi (2023). Study on how to use ‘generative AI’ in Writing Education for College students, The Korean Journal of Literacy Research, 14(5), 269-293.

Sang-eun Lee, Min-young Ku, Ye-jin Kim (2024),  Teaching Methods Using Generative AI : Writing, Programming, and Project-Based Learning (CTL Teaching Tips #45). Seoul: Sungkyunkwan University, Center for Teaching and Learning Innovation.