The featured image above is the AI Coder project team, comprised of myself, Sophia Cai, and Alec Temes as imagined by Midjourney v5. Our unique R&D project serves as an extraordinary opportunity to delve into the new and revolutionary field of AI-assisted development. Our project will result in three versions of a “Mobile Treasure Hunt” game for both iOS and Android platforms, each developed using AI tools like ChatGPT and Github Copilot at varying levels of involvement. By documenting our journey, we hope to inspire others and provide valuable insights into the capabilities and limitations of AI in software engineering.
In order to keep some uniformity to make comparisons for our research, each application will have three main agreed-upon screens. We will each build in the following screens: a Start Screen to pull users in, a Login Screen, and a Home Page conveying the two main features of the game: create your own treasure hunt or download a treasure hunt. After these three main screens, ChatGPT will lead the way and inform us how to build out the rest of the application(s).
Each of us will be applying AI tools to varying degrees as well while incorporating Midjourney for asset generation:
- I will be developing the project architecture mostly “by hand.” That is, I will write most of the code and use GPT-4 to resolve bugs and refactor code.
- Sophia will use ChatGPT intensively for the project architecture. She will use ChatGPT to completely scaffold the project structure, both large and small.
- Alec will use both ChatGPT and GitHub Copilot in their tool chain for suggested function scaffold code and refactoring.
We chose these three approaches so that we can research the efficacy of each approach in implementing a mobile treasure hunt app. ChatGPT and GitHub Copilot are our main AI weapons, with some of us dual-wielding both! Our team chose these technologies because they are revolutionizing business as we speak (Hayhurst, 2023).
Some major pros of using ChatGPT and/or GitHub Copilot will be that it will save us time when setting up the boiler plate code to get the base app running. Other pros are that it will make us “Super Googlers” by resolving bugs and refactoring code using GPT. I believe some cons we will encounter is that the code will not always be always correct. ChatGPT is not infallible and neither is GitHub Copilot. Therefore, we must be diligent in our efforts to review AI-generated work.
Overall, we are extremely excited to implement our project and learn a lot this term. We’re looking forward to a successful product while embarking on this journey to become code ninjas.
Thanks for reading! 👋🏾
“a coding ninja working at his computer station, cyberpunk, artstation, 8k octane render, unreal engine, very detailed, concept art, realistic, masterpiece, sharp, rule of thirds –ar 9:16“
Chris Hayhurst, contributor. (2023, February 23). How generative AI tools like CHATGPT could revolutionize business. Dell. Retrieved April 14, 2023, from tionize-business/
The ninjas are in the building 🥷