Our R&D project is to develop an app from scratch through the assistance of AI and recording the recording the results. With my two other group members, Alec and Tamarsh, we will individually create our own versions of a “Mobile Treasure Hunt” game for both Android and iOS platforms and use AI tools like ChatGPT and /or Github Copilot to aid us in this creation process. There are no hard and fast rules in terms of our app specifications (besides general structure and function) or level of AI use, allowing each member to develop their app and interact with AI at a more organic level. At the end of this project, our goal is to have gathered valuable insights into the benefits and drawbacks of using AI in software engineering, and hopefully guide others interested in this development process.
All members will be interacting with the AI at differing levels as well as utilizing generated images from Midjourney. In descending order of AI usage, our projects at a glance look like:
- I will heavily use ChatGPT for project architecture, feature creation, and code debugging. My application will use the Flutter SDK which uses Dart and it will be made for iOS platform. I’ve chosen these options because I took a mobile development course last quarter that used this, so the development process is fresher in my mind. However, if ChatGPT does not recommend this SDK for the type of application we are developing, then I will switch to whatever is suggested.
- Alec will be using ChatGPT and Github Copilot moderately. AI will be used to generate scaffold code and to refactor.
- Tamarsh will only use GPT-4 for code debugging and refactoring. Most of his code will be written solely by him.
With our AI usage level as the main differentiating factor between our three versions, we can investigate what the most optimal AI usage would be and why.
Our AI tools will be either ChatGPT or Github Copilot, or maybe even both! With “AI” and “ChatGPT” dominating the current cultural zeitgeist, these tools have the potential to disrupt how people work (unless it replaces us which it might…), take over common processes, speed up research, educate efficiently etc. For our development process, the possible benefits of AI will probably be generating boilerplate code as well as resolving bugs. The drawbacks would be incorrect assistance provided by AI or unhelpful advice. We will be documenting this all the while to capture key insights, and to hopefully end up with an amazing application.
Leave a Reply