Reverse Turing

Reverse turing, an oneiros minor event under MUJ ACM. The event had 3 rounds:

1. Generative AI - The participants were showed few 3 images of various difficulty, they would have to recreate the image using their own prompts and tools.
2. AI Fued - Similar to the Family Fued, the partipants would have to guess the answers for AI related questions.
3. Reverse Turing - Omegle like platform for one-to-one conversation with a set limit of 150 seconds. The participants would have to hold a conversation on the given topic (shown on chatroom UI) and based on the conversation, the participant would have to guess if they were talking to a human or AI.

NOTICE: This repository is the codebase for round 3.

Installation guide

1. Install Python.
2. On your terminal run

cd reverse-turing

# install the required packages
pip install -r requirements.txt

3. Update the creds.env (very important step).

# Database schema is given below please create a user accordingly
mongo_url = "MONGO URL e.g. mongodb://localhost:27017/ or altas url"

# This is what will be used for JWT token and authentication.
jwt_secret = "YOUR JWT SECRET"

# openrouter is the provider used for the AI. llama 3.3 70B
APIKEY = "openrouter.ai API KEY"
APIKEY1 = "openrouter.ai API KEY"

4. Run

py main.py

UI Screenshots - User

Login

Dashboard

Leaderboard

Match making

Chatroom

UI Screenshots - Admin

Login

Dashboard

Chat Log

Admin Abuse

Create User

Ban Users

Funny Conversations

Some of the few funniest conversations I found.

Database Schema

This consists of the collections required for your MongoDB. Create all the required collections in your DB.

users collection.

{
    "_id": "username",
    "password": "password",
    "token": "JWT token with {'username':'...', 'type':'user/admin'}",
    "matchmaking": true, // if true then the user is actively looking for a match
    "score": 20, // points to show on leaderboard. 5 for correct guess, 0 for wrong
    "type": "user/admin", // admin will be able to login in ADMIN portal.
    "lastpoint": 2314234532423, // epoch for last time user scored. leaderboard stuff
    "banned": false, // if true, the user wont be able to login or match making
    "previous": "AI/someone elses username",
    "judged": ["lsdyhsakldjashj"], // uuid for chatroom that were already judged.
}

Add an admin account manually.

{
    "_id":"root",
    "password":"root",
    "token":"CREATE TOKEN WITH YOUR OWN JWT TOKEN",
    "matchmaking":false,
    "score":0,
    "type":"admin",
    "lastpoint":0,
    "banned":false
}

chats collection.

{
    "_id": "UUID for chatroom",
    "messages": ["format used everywhere in openai [{...}, {...}]"],
    "user1": "username of user1",
    "user2": "AI/username of user", // AI will always be user2
    "time": 9328742983749, // epoch time
    "active": true, // true means active else otherwise
}

reports collection. (no use)

{
    "_id":  "random generated report ID",
    "reportby": "reported by username",
    "reported": "who got reported username/AI",
    "message": "the message which was reported",
    "time": 97231649823, //epoch time converted to IST
}

adminlogs collection.

{
    "_id": "action ID",
    "admin": "admin username",
    "action": "banned XYZ username",
    "time": 8721638712, // you get the point
}