iRobot

Our project is called iRobot.

Inspiration

Our inspiration comes from interacting with LLMs, Artificial Intelligence platforms, and chat bots in our daily routine. We often take these kinds of technology for granted and often devote our time to getting resources from these platforms. Sometimes we ask it life advice, studying tips, and to make useful agendas. Although, through these continuous interactions, can we really differentiate AI with ourselves?

Additionally, we wanted to create an interactive experience for students and educators to be able to realize the harmful effects it's doing to our critical knowledge. Can we truly recognize a difference or are we slowly becoming AI ourselves?

What it does

Participants will be given the same question and be asked to reply to the question. There will be a time and word limit to ensure everyone gets the same amount to respond. After the time limit is up, everyone's responses will be shown on screen and each participant can "point" to whose response is AI. They can even attach an explanation as to why they believe it's an AI response. Although, this isn't the official vote yet since there will be three more rounds of "pointing" to give the participants a chance to win. After the three rounds are up, the players can officially vote who the imposter is. If the majority of real players guess correctly, the players win. If the AI successfully escapes voting, the AI wins. If there is a tie between the participants, those players can respond once more to a question. They will be able to "point" and vote again.

How we built it

We built it using Vite, React, and Tailwind CSS since these were tools we trusted with our experience. Vite helps us speed up the development of our project, getting a better start, and has a consistent dev environment. React is a great library to use when developing games with multiple kinds of actions. This is useful for managing what the participants are capable of doing and how the UI is viewed. Tailwind CSS is an easier way to apply conditions, design it properly, and be able to apply the styles to our previous code in React. Our real time engine was Socket.io since it keeps up with the inputted communication between players and ensures the game is in sync with the participants. If the game is behind, this will make players feel disinterested in the game and our purpose of spreading knowledge of AI will be unsuccessful. Finally, we incorporated a Gemini API to make sure we had an actual AI participant responding to questions. This advances the logistics of the game, which focuses on finding the imposter and the human.

Challenges we ran into

We ran into issues with integrating our parts. Particularly in terms of the AI API and the backend, as they were done separately. Eventually, we were able to move on and fix the issue. his helped us develop skills in documentation and troubleshooting. Another challenge was managing the API credits we have since it was tough to get through the testing phase on our project. We tried to be more strategic when engineering our prompts and ensured the AI was still responsive. This limited credit budget was tough to get around but it helped us develop our problem solving skills for the better.

Accomplishments that we're proud of

We're proud of completing our game within the span of the 24-hour hackathon event at California State University, Fullerton. Our team spent hours trying to troubleshoot, planning technically, and making sure we had the resources to explore our project further. Of course, it was a major challenge to keep still and focus on our game for a whole day but we got it done in the long run. We're especially proud of our prompting and the smoothness of the real time sync of the game. This is what makes an idea into a reality which is the magic of computer science. We wanted full functionality for participants and that's what was accomplished.

What we learned

We gained experience in team collaboration getting over hurdles. We practiced our communication and we also learned the importance of authentic "human" communication and how it's not supposed to be perfect. This reflects in our project. Our responses are unique and quite irregular which is a drastic difference from AI. We had to develop a further understanding in Git workflows and figuring out the conflicts that resulted in errors in our code. We really understood the importance of having a structured framework surrounded by frontend and backend. We all teamed up and worked together to do our parts which contributed to an organized agenda throughout the two day weekend.

What's next for iRobot

This game challenges educators and students to inspect digital communication and spot detailed patterns in AI responses. When participants compare human thought responses to robotic constructed answers, students are able to really take in an observation of a real life problem. This interactive, entertaining game can help students and educators spot AI behavior more easily and critically observe what's in front of them. In a society constantly changing due to advances in artificial intelligence, we need to use our own cognitive agility to "point" at the algorithmic footprints.