Eric Chae

AI in Software Engineering Education: A Student’s Perspective on Usage, Impact, and Responsibility

09 May 2025

Introduction

AI nowadays is used everywhere in education, whether integrated in class or students are using it for their own personal education, and it has quickly become a staple for most. Especially in Software Engineering, developers and programmers are using it every chance they get. It’s a useful tool that is being utilized daily by them. Since starting college in August of 2023, I have utilized AI as much as I could — mainly ChatGPT — and recently in my Spring 2025 semester for ICS 314, I branched out to using GitHub Copilot and DeepSeek. All three of these different AIs are useful and help streamline the process for me in coding, as it can be a very long process to make programs. I had the most experience with ChatGPT, but in time I found DeepSeek to be more reliable for my coding problems, especially more complex ones that ChatGPT would get wrong often. DeepSeek would be able to spit out the answer I needed in less time.

Personal Experience with AI

  1. Experience WODs e.g. E18: Most of the time while doing these I would try my best and start them at first and go as far as I could, then use AI to help finish the rest of the assignment. But at times, if I didn’t understand it at all, I would start with AI from the start.
  2. In-class Practice WODs: Same process as with the Experience WODs — I would start them to the best of my ability from the instructions, then ask AI to help finish it for me.
  3. In-class WODs: The first few WODs we did, I would do most of it by myself then ask AI to finish and fix my code according to the guidelines required. But as the semester went on, I heavily relied on AI to complete the whole WOD for me to guarantee I could finish the WOD on time and not get a DNF on my WOD card and get only 10 out of the 100 points for that WOD.
  4. Essays: For every essay, I only used ChatGPT to check grammar and spelling mistakes as I write them straight in GitHub instead of somewhere like Google Docs that does that for you already. For the design pattern essay, I had it help me brainstorm ideas as I wasn’t familiar with the concept.
  5. Final project: For our Final Project, I used a mix of ChatGPT, DeepSeek, and GitHub Copilot to complete and finish all of the coding tasks I was assigned for our project to ensure we would meet the deadlines on time with a completed product.
  6. Learning a concept / tutorial: I would only use it to help me get a better understanding of concepts or summarize what was being told if it wasn’t quite clicking for me in my brain.
  7. Answering a question in class or in Discord: I would use AI in class at the instruction of my professor to do so for a discussion, but otherwise I didn’t use AI to answer any questions.
  8. Asking or answering a smart-question: I never used AI for these as often they were simple questions or a question that AI could not give me the answer I needed, and I barely answered anyone else’s questions as I didn’t interact much with the smart questions channel. But when I did, it was something AI could not help with or was unnecessary.
  9. Coding example e.g. “give an example of using Underscore .pluck”: I never used AI for giving me coding examples as I never needed them.
  10. Explaining code: I would occasionally use AI to explain code if it was complex and I didn’t understand the functionality, but I could understand most simple or normal things.
  11. Writing code: I used ChatGPT, DeepSeek, and GitHub Copilot all the time to write code as it streamlined the process of doing so and was very useful and convenient. Whether it was for homework or an in-class assignment, I used it.
  12. Documenting code: I never asked ChatGPT to document my code during this class as it was often never required or asked, and if it was, it was simple documentation that I could do.
  13. Quality assurance: While using AI, it was never perfect 100% of the time. So if I got an error while running or compiling my code, I would ask ChatGPT or DeepSeek to resolve my errors.
  14. Other uses in ICS 314 not listed: I didn’t use AI in any different way that wasn’t listed above.

Impact on Learning and Understanding

AI has definitely helped in my learning experience as the main thing it has done for me was complete things that I could not or do things at a much faster rate than I ever could. While it has benefited me a lot, it has also made me lazy and reliant on AI to help me with coding assignments all the time. So I have not really progressed that much in any of my skills significantly. Maybe my comprehension skills have gone up a little, but besides that, nothing has improved. Overall, AI has enhanced my understanding of software engineering concepts, as I was pretty lost at first in the class without it, and it helped me through understanding the software engineering concepts we were taught.

Practical Applications

I personally don’t know any real-life situations AI has actually been used, but I have used it numerous times to give me answers and help me outside of ICS 314. In other classes, if I don’t understand a concept, I would use it to break it down and explain it. I have also used it for real-life things. For example, if I asked Google about where a store is in Ala Moana — specifically what floor it’s on — and I can’t get the answer I need, AI can help search the web for me and it would give me the exact answer I am looking for.

Challenges and Opportunities

While using AI throughout this course, it has not been perfect — as nothing is. I would put clear instructions, even adding on to the already clear given instructions, and ChatGPT would give me faulty code that wouldn’t work. Even tweaking and revising my prompts to it for 40+ minutes would either make it worse or do basically nothing to fix my errors. But when this would happen, I would put in my problems and the code into DeepSeek and it would immediately spit out correct code solving everything I needed. For further integration of AI in software engineering, maybe give examples or talk about what you should and should not tell AI to do specifically when prompting it for classwork/homework to be efficient and correct.

Comparative Analysis

Traditional methods such as lectures and physical note-taking, or YouTube videos even, I believe are the best way to learn — at least initially. Nothing will replace this, I think, and it will always be a reliable source of information and knowledge. However, AI is best used for reviewing and quizzing. If you have done every kind of review and quiz that has been given to you, you can ask your favorite AI to help with quizzing and testing your knowledge on concepts. AI is also best used for maybe breaking down information you don’t yet quite grasp. You can tell it how you understand it, then it will go based off of that and help you understand the how and why. It will never replace learning fully, but it is useful in understanding.

Future Considerations

I think AI in future software engineering education will be very useful and vital — but with some revisions. If it’s possible, limit the use of AI. Don’t let students say whatever they want into AI; only let students use it for most things, but don’t let them do everything with AI. Also, strongly encourage doing everything once by yourself, then use AI. A good way of doing this is maybe being flexible on deadlines or advising students to do their assignments early so they have plenty of time to try it on their own, then use AI to complete it and understand their tasks. This is because AI should not be used to completely replace their entire learning. It should only assist — even if it assists a lot.

Conclusion

Overall, AI is a great tool that should be used everywhere, just with limitations. AI must not replace one’s entire learning — it should aid and assist in learning things for Software Engineering. It can be used as a crutch to do everything for you, because then you will learn nothing, which is not beneficial at all. Hands-on learning should always be on top, and then using AI to understand things you don’t are the times it should be used. For students who feel too pressured or time-crunched to try and learn software engineering concepts, a good way to relieve that is to try and give ample time for them to take their time and be slow and go at their own pace so they can really grasp the concepts.

This essay used ChatGPT for spelling and grammar mistakes.