Inspiration
There is an increasing amount of misinformation with AI online, especially political misinformation. An extremely large number of people use Social Media, and it is important that people can tell AI misinformation from something real. Most of the time, the stakes are not that high, but the entire country of Ukraine could have fell if they believed the deepfake that was propagated by Russia in March 2022.
What it does
It is a chrome extension that you can press while you're watching a video or scrolling any social media. It allows you to screen record what you're watching or take screenshots, and it will analyze the video using crosschecking with real events and AI detection. It will then give the analysis on why it detected it as real or AI, pointing out things that humans might not see. It could also recognize a real life event.
How we built it
We built this project through a mix of reverse engineering. To start, we asked Claude Max and Gemini on scripts in order to pinpoint a proper design on our chrome extension. Then, the rest of the project (taking the majority of our time), was spent prompting Loveable over and over again, perfecting the code and debugging everything from api's to frontend glitches. Also, we managed to implement an older yet functional OpenAI API key which does most of the heavy lifting in the AI detection of this project.
Challenges we ran into
The main challenge we encountered in this project was the actual backend vibe-coding part of this project. We had to prompt Loveable in a very peculiar manner which led to frustration, however- over time, we learned how to prompt better and therefore out responses and outcomes became better. Over time, we had 31 versions of our chrome extension, showing our consistency in this product.
Accomplishments that we're proud of
The main and big accomplishment we are proud of are staying consistent and grounded into the creation of our product over the course of 24 hours. We learnt that when us 3 work as a team, we can achieve such great things. We also learned how to create and manage a product that could possible be scaleable.
What we learned
We learned how to work as a team over the course of a day which improved our skills to not only communicate but collaborate too, also- we learned how to use new Loveable as a new tool, and it seemed like a very useful and efficient coding agent for us. Finally, we learned how to create a presentation for our own product which shows off our soft skills.
What's next for AuthentiScan
For AuthentiScan, we are planning to improve its general abilities and after that period of time, we will be ready to publish our extension to the google store, we can and will advertise it through social media and interactions and connections(In person or online).
Built With
- chatgpt
- claude
- css
- gemini
- grok
- javascript
- json
- loveable
- newsapi
- openai
- python
Log in or sign up for Devpost to join the conversation.