AI Safety Squad

Inspiration

We all had our own contributions to the creation of our idea; An educational game seemed to suit all of our interests best, allowing us to explore cybersecurity and ethical issues in the future of AI while marvelling at the innovative new-comings of machine learning. In addition, most of us could learn a new skill: game development in Unity. All of it came together as we were able to think more creatively than we would have if we hadn't made a story-based, educational project, rather than hyper focusing on technical aspects of the project. We were more inclined to have a better balance of creativity with technicality than to be

What it does

The game takes place in a sci-fi, data filled world, representing the datasets and machine learning models that contribute to the creation of an AI model. The main character, a cute little piece of data, is tasked with "fixing" the world it is in, by playing minigames that solve the issues with the data.

• The first minigame is meant to represent cleaning data, looking for "bad" data to destroy, and "completing" the incomplete data. This is a timed mini game, with the goal of destroying 10 data and completing 10 data within a minute, chasing after the enemies to fix them.
• The second minigame is meant to represent data privacy and avoiding the use of sensitive information in training AI model. This is done in a ChatGPT themed space invaders style.
• The last minigame (incomplete) is meant to teach other considerations that AI developers must keep in mind with regards to ethics. The main character is taken into a dinosaur runner style game, where it must collect the hearts that represent the ethics concerns that are taken into consideration, symbolizing these moral values are kept in mind when building the AI. This part of the final product is not fully finished, so would need further developing to fully serve its purpose.

After these three minigames are completed, the AI model is ready to be trained and developed, and the main character has saved this doomed world, allowing it to contribute to the innovation that the future of AI holds.

How we built it

Each person on the team took on a task. We broke them up as such:

• Minigame teleportation system, main map, and gameflow
• Asset/Graphic Design, main menu
• Minigame 1
• Minigame 2
• Minigame 3 (?) (for whoever finished their part)

Challenges we ran into

As most of us were not familiar with Unity, a lot was learned as we went along, and many times from mistakes. Some challenges we ran into were:

• Collaboration on a Unity Project with more than 3 people is complicated, so we had to use a git repository for only assets. We each worked on a different minigame/aspect of the game, and brought all of the scenes together on one computer at the end to complete the final product.
• Half of our team had little to no experience with game development in Unity, so much of their time was spent learning, and things moved a little slower as they got used to the structure and workflow, constantly debugging
• We started the project confident we could make it aesthetically pleasing and fully complete, especially with a team of 4 people. But, we only had 3 team members present on the second QHacks day due to unforseen circumstances, making the development process lengthier and more hectic for us.

Accomplishments that we're proud of

• We made something we like! We came to QHacks not knowing what to expect, and we ended it feeling accomplished. For our first Hackathon, we feel pride in the game we made, especially considering how much of it we learned on the spot and have now added to our skillset. We learned a lot and produced some impressive logic within our minigames despite being new to Unity. The fact that we have working games after spending ages debugging, wondering if it would ever work, is a huge accomplishment.
• We think the graphics in our game are quite visually appealing, sleek and simple. This gives them their charm, and they make the game look put together even if the game is not fully finished. ## What we learned
• Setting priorities within tasks is imperative when on a time crunch. We knew we wanted good graphics to add appeal to the game when our minigames would start off simple, but also knew when to switch gears from asset design to coding.
• Not everything will be perfect. It is okay to leave things to fix later if they are not crucial to the functioning of a program, and it is also okay if there isn't time to get it done eventually. The point of a hackathon is to do what you can, to search your mind at its greatest depths and produce creative ideas, not a full stack fully functioning application.

What's next for AI Safety Squad

We had always planned to add multiple minigames to place greater emphasis on more aspects of ethics and cybersecurity concerns of AI, so we will definitely expand the amount of tasks the main character has to go through to make the AI a good, valid model.

We would also like to add more complexity to the whole game, both to expand our own knowledge of Unity further, but also to increase the game's quality and refine it to a more finished product. We would do this in terms of graphics by adding animations to make things more visually pleasing, and also on the game logic side by making more interesting levels that require more time and effort.

Lastly, we would like to solidify the story of the game and present more information about the backend of AI to players to make the game even more educational. We would like to make it as useful as possible to those who may not be well versed in the future of AI in relation to cybersecurity and privacy.

Built With

• c#
• unity