As a full-stack developer at Techspire, and team lead for the full-stack team, my job involves staying on the cutting edge of technology. I’m constantly looking for new things to learn and new challenges to tackle. This drive led me to participate in Techspire’s first hackathon, where the idea was not to code as much as possible, but to let AI handle most of the tasks. This is my tale from that hackathon, of an encounter between AI and human expertise, and how it made me reflect on the current state of AI in software development.
challenge accepted!
The task assigned to us at the hackathon was to create an application for a company to rent bikes to others. As an Angular developer, I was tasked with creating the front-end for this application. However, I decided to venture out of my comfort zone and try Flutter. The goal was to let AI do most of the work. So, I created an OpenAI API key, downloaded AutoGPT, and set it a few goals: create a new Flutter application, make the application in the style of the techspire website, and set up a basic sign-in page.
autoGPT to the rescue?
I let AutoGPT run for a while, but after about 20–25 cycles, it was clear that it wasn’t making progress. It seemed to be stuck in a loop, and I was unimpressed. I thought perhaps my goals were too strict or too specific. So, I decided to simplify them and see what AutoGPT could come up with. My new set of goals was to set up a new Flutter application and create a basic sign-in page. Unfortunately, even these basic tasks proved too much for AutoGPT. It was disappointing, to say the least.
Time was ticking away, and I had to make a decision. I decided to go back to what I knew — Angular. And, interestingly, this experience made me appreciate the speed and efficiency of human expertise. Despite the advances in AI, it couldn’t match the pace at which I could develop a basic Angular application. It made me realize that, at least for now, deep expertise in a field can’t be replaced by AI.
conclusion
My experiment with AI wasn’t entirely unproductive. We used ChatGPT to come up with some snarky, sarcastic catchphrases for our application, which we named “Swap Bike”. One of my favorites was “Swap Bike because renting shoes is just weird.” It was a fun way to add a bit of personality to our app.
Despite the bumps along the way, I enjoyed the hackathon. It was an opportunity to try new things, and even though my experiment with AI didn’t go as planned, it was a learning experience. I’ll continue to explore AI tools, and perhaps in the future, they’ll be more capable. But for now, I’ll stick with Angular, and continue to hone my expertise in it.