My introduction into AI was in my first year of community college. My understanding literally came from the word itself - an artificial knowledge base that mimics what human knowledge can produce and process, this time by a machine mind. Artificial Intelligence should be an addition to the Seven Wonders of the World because of its marvelous capabilities.
Just like humans, they learn and adapt to recognitions in patterns from data however a difference is that they do an excellent job at this. But when I imagined AI, I was thinking about those big bulky robots with multiple sensors like the ones in Industrial Automation or in Healthcare. What I was not expecting was that a product of Artificial Intelligence would be more within reach and not a far fetched idea that I needed a fancy lab to access. More importantly, it has had a profound impact on my education especially in my ICS 314 classes.
Which brings me to my usage of ChatGPT. I began using this tool last semester when I was in the throes of taking 6 classes. I was particularly struggling with making summaries of weekly lessons because there was a ton of content to comb through either by reading or watching the lecture videos. I stumbled across a tik tok that recommended ChatGPT to help with studying methods such as summarizing a topic. So I took it and ran with it. In its haste, I should have stopped to carefully ponder its uses and effects which is what I talk about in this essay.
In regards to the WODs in the modules, my use of ChatGPT was minimal because the resources and the videos were adequate in explaining the material. For some lessons I did ask for additional examples of the code. For example, the resource of the underscore library had one or two examples given for each function that were a bit hard to understand at first. So I asked ChatGPT to provide more examples and then I was even more baffled when it gave me its output. It was then I noticed it was compacting the functions and this provided a more in-depth understanding of using functions as return values.
I am glad that I attended lots of the practice WODS with the teaching aids. I used chatGPT during these sessions with the TA’s to code my solutions and then later compared what it gave me to the TA’s solution. I did so by asking ChatGPT to explain the code it gave me and then gave the TA’s solution to explain their code. I took down notes of its differences and usually the TA’s solution was more intuitive while ChatGPT was more ‘quicker’.
For the WODs focused on Javascript Functional Programming modules, I was heavily reliant on ChatGPT to wire the solutions. The reason being was that there were so many functions to cram so I would ask it to give a relevant function for the WOD’s task.
The heavy use was the same for when we went into the UI Design modules. I would take a screenshot of the website and feed it to chatGPT to then make a website according to the perimeters of using such as using bootstrap or react native stuff.
As we got into application design, the use of Chatgpt for the different templates dwindled as it became specific and each problem was unique. Chatgpt was able to give a general idea of where to start but you still had to make the effort of exhausting their options to find the right direction.
After hearing from Professors who teach the English, Social Sciences and Humanities caution students on using any form of AI to write our essays, the fear it instills has held me back from asking ChatGPT to write an essay as I worry that it would be flagged as plagiarism. Although I do ask ChatGPT to write a layout of how my essay should go and it gives a good structure and the examples it gives helps me to contextualize the topic.
For our Final Project, at the time this essay was written, I was mostly assigned tasks that involved user interface and design so naturally Figma became my best friend. I would design what the websites would look like for the different pages, take a screenshot, then feed it to ChatGPT to help give me a start. Then I would have to tweak it to make it look close to what I designed. I would say it gave me a 40% headstart and I had to do 60% of the rest.
I should mention that ChatGPT also contributed in guiding us what we should put in each webpage that we like. For example, I had a resources/guide page and since I did not know much about insurance (I come from a country with universal healthcare), instead of sifting through a multitude of websites to get ideas, I asked ChatGPT and it offered a detailed section by section items for my page. I sought similar help for the other 2 pages. In addition, our team logo was generated by AI.
When I was confused about the React-Bootstrap components, I would ask ChatGPT to modify the code for me. I recall when I had errors along multiple container elements within the same JS arrow function. I provided it to ChatGPT, asking what was wrong with the code and it showed that I was missing enclosing tags. I was about to embark on a roundabout of adding colons and parentheses at random places to try and fix the error but ChatGPT saves the day.
I didn’t use ChatGPT much to learn the materials in our modules as I explored the different resources provided. In addition, some of the explanations that ChatGPT provides would vastly differ from what was given in our lessons. In addition, since most of the lessons and the code templates were proprietary to the school, ChatGPT lacked the relevance when explaining stuff and was mostly general stuff.
I did not use ChatGPT for in class questions or Discord as I barely raised questions and when I did, they were basic ones that did not warrant the help of ChatGPT.
ChatGPT does a good job in being a diagnostic service when I ran into problems with Meteor, the Mongo Database and the Deployment process. I remember this specific problem where when I ran the command meteor npm run start and it started a local web server but then it crashes with an unexpected Mongo exit code.
I asked ChatGPT about this and it gave a list of troubleshooting options. I attempted the first option of updating Node js and the version of MongoDB. Lo and behold it solved it! Before asking ChatGPT, I did skim over some websites that talked something about corrupted Mongo files and gave some weird looking command line stuff to try and resolve it but that led to a dead end. It saved me lots of time.
While the benefits of ChatGPT are somewhat rainbows and sprinkles, I personally did not properly view ChatGPT as a learning tool. For me, it was more like a shortcut tool that helped me to achieve WOD’s within the specific allocated time. As I am in the phases of the Final Project, it has come to my wretched attention that while ChatGPT may have helped me out at that time, in the long run it was creating a slippery slope for my memory retention abilities.
I find myself now regularly going back to our modules for a refresher course and this is eating up my time when doing tasks for my final project. When I was going through the modules, I would go through the readings and other resources once or twice and take attempts at activities lightly because I was relying on ChatGPT to help me out when needed. It is now being realized that this is bad practice. What I substituted for ChatGPT’s help was the learning process that would otherwise seared the content into my knowledge base.
But the ChatGPT crutch I am leaning on has not broken and I am not completely crippled. What is funny is that when I find holes in understanding Bootstrap or Meteor, it is ChatGPT who could pull me out and help explain where the knowledge shortfalls are. It helps me to fill things in where I last remember it. It’s a double edged sword. I am getting immediate answers but it’s limiting my ability to explore solutions. Yet for now, due to time constraints of the WODs and the project milestones, it’s the benefit of the immediate answers that is attractive.
Since I am a Pacific Islander, I shall identify some practical applications of AI in the Pacific region, particularly focusing on my home country Fiji. AI is an emerging field and shows promising aspects although it comes with some challenges. While development is still in its early stages, public and private companies have shown interest and the Government has launched a strategy to include AI in its technology development plan.
A good example of a practical application is the use of mobile banking. In Fiji, several banks like Bank of the South Pacific, ANZ, Westpac and several telecom companies have started using mobile banking and mobile wallet technologies. I would assume that these financial institutions would use AI to analyze user data and suggest personalized financial products. They would also use AI to provide insights of market trends and make predictions.
Outside of the Financial domain, AI would be useful in the top two industries in Fiji - tourism and agriculture. Just like for the banks, AI can help optimize travelers’ experience by offering personalized recommendations for touristy activities by analyzing individual traveler preferences and past behavior. As for agriculture, AI can analyze data from satellite images, drones, and sensors to monitor crop health, soil conditions, and weather patterns. Interestingly I watched this documentary of a Singaporean farm where AI enabled precision agriculture, where resources like water, fertilizers, and pesticides are applied optimally based on the specific needs of each section of a farm.
I think the biggest challenge I encountered in using ChatGPT for this course is the skill of making good prompts. Majority of the time, the questions I asked ChatGPT for help were vague, which I then had to additionally clarify or rephrase to get the answer I was looking for, which is time consuming.
Poorly worded questions can lead to ChatGPT to misunderstand the intent or context of my question, resulting in answers that didn’t address my actual needs or concerns. I had my poorly worded questions moments and then there were my overly worded questions moments that overwhelmed the AI, causing it to generate disjointed or incomplete responses. Both scenarios lead to a frustrating back-and-forth interaction to get a clear answer.
To better integrate use of AI into software engineering classes, I think there should be a module that dives into perfecting your prompts. I think this would have students putting more effort and trust into AI as the frustration of consistently receiving responses that don’t meet their expectations or needs is reduced.
What I liked about this software engineering class is how, in my viewpoint, it divides from the traditional classroom learning approach. I was surprised when the instructor told me that using AI for your module experiences and WODs were ok when it felt like a cardinal sin with how my other classes preached it to be.
The biggest difference is the immediate answers and feedback that ChatGPT offers. I did not have to bother our instructor at evening times or during the weekends when my code was not making sense. I did not have to wait for my instructor to respond hours or days later to a question I asked. Furthermore, I did not have to regularly book office hours when I could just consult ChatGPT to explain or solve it for me.
Additionally, the use of ChatGPT to summarize topics is advantageous to having to read through pages and pages of textbooks, watch multiple youtube videos and comb through lesson notes - methods we are used to in the traditional classroom setting. Learning becomes personalized. I choose the content that I find difficult in learning and have someone explain it to me instead of sifting, and sifting, and sifting…
I think this is a good question for me who is focused on being an Educational Technologist. In the future, AI can be used to keep learning management systems with educational content that is current, integrating the latest trends, activities, insights and technologies, making the curriculum more efficient. In addition, it can immediately grade students assignments and gives instant feedback so that the students can go onto other activities.
To conclude, AI is beneficial to a Pasifika student like me who comes from a diverse educational background having gone through different learning styles. Engaging AI tools to Pasifika students can help bridge educational gaps and cater to specific learning preferences. Moreover AI-enabled remote learning platforms can provide access to global resources and expertise wherever the student is learning from!