-
player made a descriptive prompt!
-
player made a vague prompt :(
Inspiration
As the use of AI increases, it becomes critical to know how to effectively use it. When using AI to generate content, the output can be different when you don’t know how to properly prompt it. Even if the content is the same, how you communicate your prompt makes a difference. We aim to equip the next generation, growing up in the age of AI, with the skills to interact with these tools thoughtfully and effectively.
What it does
Duck AIventure is a game designed to teach school-aged children how to effectively prompt AI. Users are given a target word and must prompt the AI to say the word. The AI evaluates the prompts based on their effectiveness and returns a score, which is then used to determine if the user can move on to the next challenge or “bug.”
How we built it
To build this code we used Python with the pygame library to create a functioning game. Additionally, we used an API key from Google Cloud to make API calls to the Google Gemini LLM from within our game.
Challenges we ran into
We ran into many issues while working on our project. One of these issues was the constant merge conflicts that resulted from us not updating our local repository frequently enough, especially when big changes were made. We tried to fix this issue by testing on three of our four computers and only pushing code on the fourth computer. Another issue was layering our images properly. Because of all the screens and toggles included in our game, it was hard to track which image was displayed at what time. Oftentimes we had problems with images and icons displaying when they weren’t supposed to. To fix this we had to make sure the screen was properly resetting each time we wanted to remove something from our screen. The final issue was setting the boundary issues. It was hard to create boundaries for the bug to prevent it from going on areas of the map where the player can’t properly see it.
Accomplishments that we're proud of
One of our main accomplishments was being able to implement GeminiAI in order to make API calls. Additionally, we were able to successfully use PyGame to make a fully functioning game without much prior experience. As a first-time and all-freshman team, we are proud to have made a functional finished product by the end of the hackathon.
What we learned
We learned a lot about the pygame library and its boundaries during this project. Two of our team members had very little knowledge of pygame prior to SwampHacks, but through research we were able to navigate the methods included to create a functioning game properly. Additionally, we learned how to implement generative AI into our code to allow users to chat directly with Gemini and get feedback on their prompts.
What's next for Duckie AIventure
In the future, we plan to expand our game and the world within it by including adding combat and animations, scrolling backgrounds so that the environment can change, threshold increase as you collect bugs, and a more solidified storyline. In addition, we plan to add a way to display scores and include a leaderboard to engage with other players. On the front-end side, we plan to make it more dynamic by adding animations to our backgrounds, creating a textbox that adjusts to the input, and a typing animation for the Gemini response. We also want to embed our game on a website to be more accessible to our target audience. This will allow it to be used in classrooms to teach children about prompting AI.
Log in or sign up for Devpost to join the conversation.