Show HN: AI Election Predictor – Test Your Voting Demographic

I've built an MVP that uses AI to predict voting patterns based on demographic data. You can try it here without signup: https://tinyvotes.com/ https://tinyvotes.com/ What it does: - Input parameters like age, gender, state, county, education, and income - AI predicts and classifies voting choice for that demographic - Global counter shows aggregated voting statistics Tech stack: - Frontend: Vite for fast development and optimized builds - Backend: Firebase ecosystem - Firestore for real-time database - Cloud Functions for serverless logic - Cloud Storage for parameter and prediction logging - Firebase Hosting for deployment - AI: Leveraging both Grok and Gemini for predicting and classification Why I built this: - Curious about the intersection of AI and political forecasting - Wanted to create a tool that could potentially offer more granular insights than traditional polling - As an immigrant, I'm fascinated by the diverse political landscape of the US - This project helps me (and potentially others) understand the complexity of American demographics and voting patterns Some interesting findings so far: 1. Not everyone in California votes for the Democrats, contrary to popular belief 2. The combination of younger age, female gender, and lower income shows a higher probability of being undecided 3. Most intriguingly, the AI sometimes predicts "undecided" as an outcome, despite not being explicitly prompted with this option. This showcases the LLM's ability to capture nuanced political stances beyond simple binary choices. I'm particularly interested in: 1. How accurate do you think this kind of tool can be? 2. What are the ethical implications of such predictions? 3. How might this impact traditional polling methods? All feedback welcome - on the concept, implementation, or potential use cases.