Inspiration
In 2021, over $500 million was reported lost to romance scams, according to the FTC. This is just in the United States, and likely only a small fraction of the total money lost, as not many people actually report these scams. Many poor students fall victim to these scams, with the scammers promises of easy financial returns. Scammers have gotten extremely detailed with their tactics nowadays (see the Pig-buchering scam), with head offices, "sales departments," and even HR psychology-manipulation manuals.
What it does
Our goal to combat this scam is to have the scammer talk to AI. The AI pretends to be a real person, and chats with the scammer. Every minute a scammer talks to an AI bot is a minute that they can't scam a real human. We set up an implementation with Telegram, so you can enter a message ID of a chat, and the AI will start responding to messages sent in that chat. The AI is surprisingly good at conversation and it is difficult to know that it is not a human.
How we built it
We implemented the AI and Telegram connection in Python. The AI used was Microsoft's Pretrained Conversational AI (microsoft/DialoGPT-large). Our front-end dashboard was built with Vue.js/Tailwind.css, and the API to connect the two was built with Falcon in Python.
Challenges we ran into
This is quite a large project for 24hrs, so it's currently a rough implementation. We ran into a lot of challenges making the program work across different OSs, but we're confident it's (pretty) stable right now.
Accomplishments that we're proud of
We're proud of building something like this is such a short time, and even though it's far from finished, we hope to improve and build upon it until we can package it cleanly into one app.
Built With
- css3
- html5
- javascript
- python
- tailwind
- vue
Log in or sign up for Devpost to join the conversation.