If you think about it, AI is still a relatively new field, yet there are already so many use cases across many industries. One of which is in education, where AI is used to teach students, and students utilize AI to teach themselves. Although there weren’t many instances in which the professor used AI to teach us this semester, there were a few times he instructed us to ask AI during class to look up definitions and such. For the other case of utilizing AI, many students in the class, including myself, used a chatbot to generate code for us. Whether it was for educational purposes or to simply copy and paste code, there is no doubt that at some point, we all used it (Most students didn’t raise their hands when the professor asked whether they attempted to solve the WODs without using AI). I used ChatGPT, Gemini, and Claude throughout the semester, and the reason why I switched between these options is that new models were being released and benchmarks were being broken by different companies.
Experience WODs e.g. E18: For this WOD, I asked ChatGPT “Can you list possible functional programming methods that can be used to implement this function?”. In this case, I found AI to be useful as it compiled information from many sources into an organized output, and knowing what functions to use helped me formulate how I should approach the problem. Note that I didn’t straight up ask for the solution.
In-class Practice WODs. In laulima, it doesn’t distinguish between “WODs” and “Practice WODs” so I’m not really too sure which one is which. I’ll use the ko-hana-react WOD, for example. I remember asking the AI “Can you center and align the text so that it is centered?”. I had trouble with styling.
In-class WODs (UI Design (Basics)): For this WOD, I decided not to use AI because of how simple the WOD was. I believe students need to study some minimal amount so that they feel comfortable completing WODs like this one without using AI at all. If you had to use AI on this WOD, then you probably copied and pasted your way through this semester.
Essays (This essay): I plan not to use AI for this essay as it heavily focuses on my experiences. Although, actually, I might scan it at the very end for a grammar check. Otherwise, the content of the essay is too personal for AI to be able to complete it.
Final project: I used AI on several issues, mostly on concepts I was not familiar with. For example, I was tasked with the AI work, and I didn’t know how to fetch the data stored in the database for my Python code to use. “How do you fetch data stored in a Postgres database via a React form and parse it using tokenization so that my AI will be able to use it as input data?”
Learning a concept / tutorial: I use AI all the time to learn new concepts. Adding on to the above example, I asked AI how to deploy Python backend code. “How can I deploy my Python backend code on Vercel so that the system that works locally will also work when deployed?”. I found out that the Python backend must be deployed separately on a service like Heroku or Render. Then the front end on Vercel must point to Render as well as Render pointing at the database hosted on Vercel.
Answering a question in class or in Discord: Most of the time, I didn’t use AI to answer questions. In our class debate, for example, all of the information I used in my argument was found through personal research online.
Asking or answering a smart-question: Did not participate in this at all, I believe.
Coding example e.g. “give an example of using Underscore .pluck”: I’ve used AI for many functional programming methods. For example, “Can you explain .map in functional programming”.
Explaining code: A lot of my coding knowledge comes from prompting GPT. I would find myself using it as a search engine most of the time. So, for example, say I explain code without using AI. But most of my knowledge comes from AI (that can also easily be found on the internet). Will this be considered using AI? I didn’t use AI when the code was being explained, but my knowledge bank is mainly based on AI. hmm.
Writing code: Definitely yes, I would bet everyone in our class generated code at least once during this semester. This example is vague but an example would be “Can you show me how to create a simple Navbar in React and style it using React-bootstrap?”.
Documenting code: Yes, I’ve used AI to document my code. I wouldn’t use it for all, but for example, I often forget how to document code, so I ask, “Can you document these x functions? Include function parameters and descriptions of what it does?”. Then I’ll follow the guideline and document the rest myself.
Quality assurance e.g. “What’s wrong with this code ” or “Fix the ESLint errors in ”. I usually copy and paste all relevant code with the error and ask, "Here is my code and the error I am getting, can you troubleshoot this?".
Other uses in ICS 314 not listed: I often repetitively use a singular AI chat for each respective topic. So in this case, I have a chat for ICS 314, and because it has history context, I asked “Can you summarize important functional programming topics and give code snippets as examples?.
I feel like I heavily relied on AI during my earlier days of discovering AI, which was during my Senior year in high school. Then I started to realize how dependent I am on it and that I am learning basically nothing. Then, when I entered college and eventually into ICS 314, I mostly use it as a search engine. Many of the WODs I tried to attempt by myself first. I also ask the AI to give me practice questions and challenge me on topics, like interview questions, and that helps my problem-solving abilities. It really depends on how you use it as a tool. Use it to cheat, then it won’t help you at all, but use it to challenge yourself, then you’re growing infinitely.
I was a cohort for a program called AI4ALL and I was grouped with 4-5 students across the country. We used AI to come up with possible project ideas, we used AI to create a schedule/milestones for completing the project, and we used AI to help us develop our project. I believe it helped us a lot, especially since we were all online. It’s more difficult for us to communicate, and the AI acted as the intermediary for us to work together and complete the project.
Although AI was quite intelligent already at the beginning of the course, it still had room for improvement. Also, that’s considering that I had access to the latest technology since I pay for a subscription. So, for those who are behind a paywall, the chatbot would be less intelligent, which I’m sure hindered them on some assignments. However, for the most part, I was pretty content on how AI was used and how much it helped me understand concepts. On the individual side, AI in education is really dependent on the student on how they utilize it. As for the teachers, I’m not really sure how they would utilize it to teach their students. Maybe they could teach students the correct way to use AI so that they won’t entirely rely on AI for answers, and instead, use it to ask smart questions, ask it to generate practice problems etc.
Traditional teaching methods offer more human-to-human relationships/mentorships and structured environments. These are known to be important for social learning and discipline. However, traditional methods fall short in real-time adaptability as the class grows larger, hence, a scalability issue as well. On the other hand, AI approaches has personalized learning like how I talked about, personal and continuous feedback, and since each student has access to AI, very scalable to larger class sizes. Good for retention and practical skills as well, if you prompt it and utilize it correctly.
We’ll be able to see things such as autonomous learning assistants as AGI is around the corner, where it learns the student’s strengths and weaknesses and adjusts learning style. It’ll also be able to analyze their code and software engineering skills while giving continuous feedback. Unfortunately, this will lose precious human-to-human interactions, where it might eventually feel dystopian. We must also consider data privacy if it’s going to learn a lot about the student.
Although AI for education is still relatively new and (maybe) the best practices haven’t been implemented yet, in the foreseeable future, we’ll see many improvements in the use of AI in education as discussed above. Personalized learning systems are definitely achievable with real-time feedback. It’s important to focus on how the students use AI. If they use it to just generate answers, then they are not learning anything. Maybe a model that safeguards text generation should be considered. For example, YouTube Kids blocks inappropriate content. Maybe a model could be developed for students so that it’s for educational purposes only. Overall, although there are exciting directions for AI in education, traditional teachers still play an important role in guiding students, providing human interactions, and creating environments that AI is not capable of doing.