Insightful: Deconstructing Viral Misinformation with AI
Project Overview:
Problem: The rapid spread of misinformation and misleading content on platforms like TikTok and Instagram makes it difficult for users to discern truth from falsehood, leading to widespread confusion and potential harm.
Solution: Insightful is a mobile application designed to combat this problem by leveraging AI to analyze viral video content. Users can simply share a TikTok or Instagram link, and Insightful will automatically:
- Segment: Break down videos into precise, claim-based segments, isolating key statements for focused analysis.
- Classify: Categorize segments as "false information," "opinion," "misleading," "verified fact," and other relevant categories, using GPT-4o.
- Explain:
- Provide clear explanations for each classification, citing reputable sources and evidence.
- Offer direct links to supporting materials, allowing users to verify findings.
- Highlight potential biases and logical fallacies.
- Engage: Offer an interactive AI Agent for users to delve deeper and clarify any doubts. The AI Agent can provide more detailed explanations and help dispel any remaining uncertainties.
Impact: Insightful empowers users with the tools and knowledge to critically evaluate online content, fostering media literacy and promoting a more informed digital environment. By providing instant, AI-driven analysis, the app aims to mitigate the negative impact of viral misinformation.
Uniqueness: Insightful simplifies video fact-checking. While tools exist for text, our app uniquely addresses video content. Instead of tedious manual transcription, users simply share a TikTok or Instagram reel to instantly analyze its validity. We segment videos, provide clear classifications for each segment, and offer AI-powered explanations, making video fact-checking accessible to everyone.
Monetization Strategy: Insightful will utilize a tiered subscription model. A free tier offers basic video analysis, while premium tiers provide unlimited analyses, advanced AI features, and priority processing. This approach ensures sustainable development and continuous improvement, allowing us to deliver a powerful tool for combating video misinformation.
Team Information:
- Dawid Drobny
- Role: Frontend Developer & AI Agent Integration
- Contributions: Created the user interface using React Native in Android Studio and integrated the ElevenLabs API for a conversational AI Agent, allowing users to interact and gain deeper insights.
- Hubert Jastrzębski
- Role: Backend Developer & AI Analysis Pipeline
- Contributions: Developed the core backend logic, building the AI analysis pipeline. This involved processing the output from Whisper using the fal.ai API to connect with an LLM, and designing a prompt to extract structured data, which is then parsed and formatted for seamless integration with the client-side UI.
- Mikołaj Zabłocki
- Role: Server & Data Processing
- Contributions: Developed the server infrastructure, implemented the functionality to download videos, convert them to MP3 format, and perform speech-to-text conversion using the fal.ai API and Whisper.
Technical Details:
- Tech Stack:
- Frontend: React Native (Android Application)
- Backend & Server: TypeScript (Server-side logic, data processing, API integration)
- AI Integration:
- ElevenLabs API (Conversation AI Agent)
- fal.ai API (Speech-to-text with Whisper, LLM interaction)
- Repositories:
Built With
- elevenlabs
- fal.ai
- openai
- react-native
- typescript
Log in or sign up for Devpost to join the conversation.