Balancing AI and Learning
15 Dec 2024I. Introduction
Looking back at my journey in programming, it’s kind of wild to think about how much has changed with AI entering the picture. I remember the days when Stack Overflow was my best friend, and I’d spend hours searching through posts and reaching out to older students for help. Now, with tools like ChatGPT and Github Copilot at our fingertips, the whole learning experience has totally shifted. To be honest, I probably use AI more than I initially thought I would, and sometimes that makes me pause and think. But hey, those tools are here now, right? So why not use them to our advantage - as long as we’re thoughtful about it and actually reflect on how we’re using them, which is exactly what I’m doing now.
II. Personal Experience with AI
Let me break down how I’ve been using AI throughout my Software Engineering, ICS314, course:
Experience WODs
For our Experience WODs, I actually saw these as a great opportunity to tackle these challenges without relying on AI. The Experience WODs were my safe space where I could really learn and understand things properly. Even though they were supposed to be timed, I gave myself permission to slow down and work through problems the old-school way - you know, actually figuring stuff out step by step without rushing to AI for answers. Sure, I often had to watch the screencasts to help me get started, but that felt more authentic to my learning style. I’d work through the WOD first at my own pace, then try to speed it up on subsequent attempts. Looking back, these were probably the assignments where I used AI tools the least, and I’m kind of glad about that. It gave me a chance to build that fundamental understanding that you just can’t get from quick AI solutions.
In-class Practise WODs
For practice WODs, I tried to keep it real and limit my AI use. These were my chances to actually test what I knew without a safety net. But even though these WODs were only for practice and not graded, I could often feel myself getting stressed If I got stuck or used too much time. Then I would peek at what ChatGPT had to say, and he usually had a lot of useful things to say.
In-Class WODs
Let’s be real - during in-class WODs, time pressure is no joke. And you probably already guessed it - if I got stressed during my practice WODs, I most definitely got stressed during my graded in-class WODs. ChatGPT became my trusty sidekick here. When I get stressed I find it hard to concentrate, and sometimes I just needed ChatGPT to help see why I was getting errors or why my code was not doing what the task description asked for. Though I made sure I understood what it was suggesting before blindly accepting anything.
Essays
For essays, AI became like a brainstorming partner. Instead of staring at a blank screen, i’d bounce ideas off ChatGPT. Still, all the actual content came from my own experiences and thoughts - the AI just helped me get the ball rolling and structure my thoughts.
Final Project
The final project was where I really leaned into using GitHub Copilot, which I had not been using before. It turned out to be a game-changer. For instance, when I needed to create new React components, Copilot would suggest the basic structure right away - all the imports, component skeleton, and even prop types. It also helped a ton with repetitive tasks like setting up our validationSchemas.ts file.
What I appreciated most was how much time it saved me when debugging. Instead of spending hours scratching my head over why my code wasn’t working or why I was getting certain errors, Copilot would often spot the issue right away or suggest fixes that actually made sense. This meant I could spend more time actually building and improving features rather than getting stuck on technical hiccups.
Learning a concept / tutorial
AI is great for learning new concepts, and this is probably one of the smartest ways I used AI throughout the course. WHat I loved about ChatGPT was how it could break down complex ideas into simple terms, and if I still didn’t get it, I could just say “Hey, can you explain that even simpler?” or “Can you give me an example?” And boom - it would try again with an even clearer explanation and some code examples to boot. When using AI you don’t have to be scared to seem stupid, you can ask as many times as you need until you really get it!
Answering a question in class or in Discord
This one’s pretty straightforward - if I knew the answer, I answered. No need to complicate things by bringing AI into it. When I felt confident enough to help out a classmate, it meant I actually understood the concept well enough to explain it myself. That’s the kind of knowledge you want to share directly.
Asking or answering a smart-question
The smart-question channel turned out to be super valuable for me! I remember this one time where I was totally stuck. I spent hours trying to figure out why one of my essays wouldn’t get posted on this github.io site. ChatGPT could not save me on this one, so instead I used chatGPT to better understand my issue so I could ask a clearer question. And I’m glad I did! Turned out another student had run into the exact same problem, and all it took was changing one line of code to get everything working again.
Coding example / Explaining code / Writing code
These three really go hand in hand in my experience. While I used AI differently for each, they all kind of blended together in my learning process. For coding examples, I’d ask for them to better understand a concept. When it came to explaining code, I would try understanding it myself first, and then use AI to fill in the gaps if needed. As for writing code, I tried to find a balance. I’d use AI for inspiration or when I was really stuck, but I made sure to write a lot of code myself too.
Documenting code
Documentation is another area where AI can be super helpful, though I realize I probably didn’t use it as much as I could have. Looking back, I see some missed opportunities - it could have helped me write better commit messages during our final project, or just helped me document my thought process when I was deep in the coding zone and everything made sense (but might not have made sense later!).
Quality assurance
This is where AI really got to shine for me in ICS314. Whether it was tracking down bugs, fixing ESLint errors, or just making sure my code was following the best practices, AI tools were incredibly useful. They saved me tons of time on the nitty-gritty details and helped me catch issues I might have missed otherwise. Looking back at the time before AI I remember one time I almost spent a whole day trying to figure out why my code wouldn’t run in my OOP Java project - it turned out I was missing a curly bracket. Today that would have taken me less than a second to fix with AI.
Other uses
The categories we’ve covered pretty much sum up my AI usage in this course. While I’m sure there are other creative ways to use AI in software engineering, these were the main ways I found it helpful in ICS314.
III. Impact on Learning and Understanding
My experience with AI in learning software engineering has been transformative in both positive and challenging ways. On the technical side, AI tools have significantly accelerated my understanding of complex concepts like React components. When I struggled with understanding hooks in React, for instance, AI provided clear, contextual examples that helped me grasp the concept faster than traditional documentation alone. However, this efficiency comes with trade-offs. While I can quickly find solutions and understand new concepts, I’ve noticed that the quick answers sometimes mean I spend less time wrestling with problems. Those moments of struggle, though frustrating, often led to deeper understanding in my previous programming experiences. I’ve had to consciously balance using AI as a learning accelerator while ensuring I’m not bypassing important learning experiences. The impact on my problem-solving abilities has been interesting - AI has actually helped me become better at breaking down problems. By observing how AI tools decompose complex issues into smaller, manageable parts, I’ve learned to approach problems more systematically. This has enhanced my overall software engineering mindset, teaching me to think more structurally about code organization and system design.
IV. Practical Applications
Beyond ICS 314, I’ve seen and experienced how AI has become an essential tool in real-world software development. One interesting example I’ve come across is Highcharts GPT, an AI-powered tool by Highsoft. Highcharts is already a popular JavaScript library for creating interactive data visualizations, but Highcharts GPT takes it further by using natural language processing to simplify chart creation. Instead of writing complex configuration code, users can just describe what they want, like “Create a bar chart comparing monthly sales in 2023 and 2024,” and the AI handles the rest. So cool! And it gives the developer the opportunity to focus their time on more complex issues in their code.
V. Challenges and Opportunities
The biggest challenge I’ve faced with AI tools is definitely finding the right balance in how I use them. Sometimes I catch myself reaching for AI help too quickly instead of trying to work through problems on my own first. When should I push through and solve something on my own, and when does it make sense to ask AI for help? It’s like having a super-smart friend who’s always ready to help - great to have around, but you don’t want to rely on them for everything.
But there are also some really exciting opportunities here. For example, AI could be amazing for creating personalized learning paths. Imagine if we had AI tools that could adapt to each student’s learning style and pace - helping with specific areas where we struggle while pushing us to think independently in areas where we’re stronger. In ICS 314, I could see AI being super helpful for providing more targeted feedback on my code, especially during those tricky WODs were sometimes I just need a little hint to get unstuck.
VI. Comparative Analysis
Looking at how we learned before AI versus now, there are some pretty clear differences. Traditional methods, like reading documentation and working through tutorials step by step, often took longer but forced us to really understand each concept thoroughly. I remember spending hours debugging my first Java project - it was frustrating, but man, did I learn from those mistakes! With AI tools, learning can be much faster, and in some ways more practical. Instead of getting stuck on small syntax errors, we can focus on understanding bigger concepts and solving real problems. For example, when I was learning React, AI helped me quickly understand component structure, which meant I could spend more time actually building useful features. The catch is that it’s easier to fall into the trap of copying solutions without fully understanding them. I’ve found that the best approach is actually mixing both methods - using AI to get unstuck and learn efficiently, but still taking the time to work through problems the traditional way when I need to build that core understanding.
VII.Future Considerations
As these AI tools keep getting better (which they definitely will), we’ll probably need to rethink how we teach and learn programming. Instead of debating whether to use AI or not, we should focus on teaching smart ways to use it. It’s becoming as essential as learning to use Google or Stack Overflow was back in the day.
The real challenge for the future will be figuring out how to teach programming in a world where AI can write code for us. I think the focus might shift more towards understanding systems and solving problems, rather than memorizing syntax or writing everything from scratch. For example, maybe future software engineering courses will put more emphasis on system design and architecture, using AI tools to handle more of the routine coding tasks.
VIII. Conclusion
Looking back at this semester, I can see that AI tools have definitely changed how I learn and work. Sure, I might use them more than I initially thought I would, but that’s not necessarily a bad thing. The key has been staying aware of how I’m using them and making sure they’re helping me learn, not just giving me quick fixes. As we move forward, finding that balance between traditional learning and AI assistance will be super important.
One thing that could really benefit future students is having clearer guidelines for using AI tools. For example, knowing when it’s appropriate to ask ChatGPT for help and when it’s better to try solving things on your own. Both skills are important, right? You’ve got to understand traditional coding methods while also mastering how to collaborate with AI—because that’s just the reality of being a software engineer today. After all, these tools aren’t going anywhere - we might as well learn to use them wisely!