This past summer, I made a little Zelda-style Gameboy game for my son for his birthday, using Gameboy Studio. It's nothing special—little inside jokes, and lots of Zelda-derivative puzzles. Made for an audience of 1. Of course, there was tons I didn't know about creating a Gameboy game, and so I used AI to help troubleshoot and learn how things worked.
And it was great! When I might have had to search through documentation for answers, I could just ask the AI, and it generally got it right, saving me lots of time.
At one point, something wasn't working quite right—some sort of visual glitch—and it wasn't obvious to me how to solve it. I don't remember now exactly what it was. What I do remember is that I immediately tabbed over to ChatGPT, and started to type out the problem and ask for help. Midway through typing the question, I figured out the solution.
I stopped and turned to my partner, who was working on grading some papers and said, When did I get so dependent on AI? This wasn't even a particularly difficult problem. I had to think about it for all of 5 seconds to find the answer. But my gut reaction was to not even try. How many harder problems have I stopped trying to figure out over the last year or so?
For context, I've been building Mac and iOS apps since 2009. Since 2012, I've been able to support myself and my family by building and maintaining a couple mildly successful apps. I didn't go to school for computer science, so I taught myself to program. My first Mac app was actually written in Applescript, back when Xcode actually had the capability to ship full apps made with Applescript Studio! Over the years, I learned Objective-C, and then Swift, and more.
All that is not to say I'm a particularly talented programmer, I don't think I am. But I know I'm capable of working through problems. In fact, it's one of the things I enjoy about coding.
So What Happens to Problem Solving in the Age of AI?
What would have happened if AI had existed when I was learning to code back in 2009? It would have been so much less of a struggle to ship something! But would I have actually learned or internalized as much? Or anything at all?
As I've said, AI is an amazing tool. It's helped me solve obscure bugs in my apps, and saved me lots of time on boilerplate stuff. But I wonder what we will lose if we stop trying to solve problems ourselves? Will we
I'm know I'm not the first person to struggle with this, or the first to write about it. And I'm the first to admit I don't know the answers to these questions. It's just something that has made me think.