Fitness Tracker

Initial Goals: Utilise the AI to generate a relatively normal website/app for the hackathon

This Hackathon is unique in the sense that the AI does a good portion of the heavylifting in terms of generating code. However, generative AI in its current state is far from perfect when it comes to coding, even more so when it can only rely on prompts made by the user. This means that I have to: Understand what language it uses (alongside the syntax), what each line of code does in the program, the AIs limitations when it comes to understanding what I want and identifying bugs where possible. To that end, for my first entry, I decided to prompt MeDo to create a simple first-person 3D game where the objective of the player was to kill some enemies by shooting at them (I also basically gave the AI complete freedom). Needless to say, the AI performed rather badly, though I believe it is due to the fact that as a user, I should be more specific when it came to what I wanted. Aside from that, the MeDo application did not appear to understand what engine (and by extension, what coding language) it should've used despite me saying Unity (a game engine that uses C#).

Big issues that I immediately noticed:

• Enemies simply phasing through the floor (lack of proper collision model interaction with the floor), but are able to attack the player
• The player is unable to shoot the enemies
• Lighting is almost non-existent
• A plain UI
• No pause feature or settings menu
• And more

I decided to pivot to making a more basic web app, as MeDo will have many issues with making 3D games by itself without significant help (modeling, basic map designs, audio, etc).

As mentioned above, I decided to swap objectives to make a webapp with MeDo that is slightly less complex and intensive on the creative processes. Since this Hackathon also promoted apps that provided lifestyle benefits I settled on making a simple activity tracker app that allows the user to take note of their activity and estimated calorie consumption across a year. There are other features that I tacked on like:

• A BMI Tracker
• Activity Streak (similar to Duolingo, but less needy)
• App appearance toggler (light and dark mode)
• Editable activity log that allows previous log entries to be deleted
• Basic login and register page, with a sign out button should the user desire to use it

While making this app, I ensured that my prompts were more vivid then just saying "make a sports tracking app" and noted down the majority of the features stated above in the initial prompt. This helped to remove most of the issues faced in the first entry where the AI was given too much freedom, which resulted in a subpar product that would need more time to fix than add on. There were some mistakes as well when it came to code generation, although it was mostly oversight from the AI that I could easily debug from looking at the code; there was an instance where the number under "current streak" did not change, even if there were consecutive days' worth of activity logged, this was solved by referring to the dates of the current day relative to the previously logged entry.

Built With

• html
• javascript
• json
• tsx
• typescript
• yaml