Gamma

A gpt powered content moderation tool to flag hate speech, misinformation and harmful activities on discord. It focuses on explainability and AI tags with moderation tools to aid in user friendliness.

Comment

Inspiration

Our inspiration for Gamma came during the Covid19 pandemic. During the pandemic, many students became avid discord users. While participating in online school we noticed a large increase in misinformation and hate speech online. A discord moderation bot is an approachable way for developers like ourselves to make a difference and combat this misinformation and hate speech online

What it does

Our content moderation bot, named Gamma, gives defined tags such as Misleading, Discriminatory and Promoting Self Harm to an openai gpt. Using these tags the gpt flags statements which may fulfill them and provides a comment for the moderator to review as well. In addition to the original message, the sender, the timestamp, the platform and a link to the original message are provided. All of this returned information is put into a table for a moderator to review every case as they would like.

How we built it

We built Gamma by splitting into three main sections. One being the front end table development using React and Next.JS. We based the template on the open source MUI (Materials UI) website. From there we utilized React and Next.JS to expand upon fit it for our use case. The second main section was the development of a database using the python Django library. The third section was the development of the moderation bot itself. We used the discord.py library and developed the gpt in the openai playground. From there we used http to send data to and from our servers to be used by the gpt.

Challenges we ran into

One of the challenges we ran into included managing state and context of data. When passing data around between three separate sections of the project things are bound to go wrong at times and it took lots of troubleshooting and debugging to make sure the data was in a useable state for each area of Gamma. Another challenge we faced was agreeing on a format for API endpoints. We resolved this by talking through it together by reaching a consensus as a team and implementing it in main.py.

Accomplishments that we're proud of

As a team, we're proud of how well we worked together. We were friends before the hackathon but had never taken on a project of this scale together. I'm proud to say that we're still friends afterwards. We're also proud of how much each of us learned during this development. We started the development each with our own distinct skills, such as one person being skilled in Django and another being skilled in React. We helped each other learn and understand new skills so we could build the best bot possible. All together, we're really proud of our final product. This is the hardest each of us has worked in a 24 hour period in our lives and we believe our hard work has payed off.

What we learned

One of the hardest unexpected challenges was hooking up the backend to the frontend. We originally thought it would be quite simple and take only a short amount of time. After trying to start it we ran into issues such as the front and back end not communicating properly, and the aforementioned data states. After refactoring some of our original code base we managed to get the front and back end working properly.

What's next for Gamma

While currently functional with just Discord, Gamma was built with expansion to other platforms in mind. There is a section to sort by platform in the front end. Additionally, with how much we've learned in the past 24 hours, given more time we could expand to platforms like Reddit and X (Twitter).

Built With

Share this project:

Updates