AI or NOT - Paris Vibe-a-thon: “Build a One-Person Unicorn” — November 15, 2025
AI Tinkerers - Paris
Hackathon Showcase 2nd Place Winner

AI or NOT

Team consisting of Boyu Niu (Inalco NLP researcher, PhD/CIFRE — LLMs/VLMs) and Abhinav Singh (HEC Paris PM, Ledger; blockchain, product/LLM strategy).

2 members

“AI or Not?” is a gamified mobile platform where users swipe right/left on hyper-realistic AI-generated videos to classify them as real or fake -similar to Tinder but for media authenticity. Users compete on a global leaderboard, earn rewards, and monetize their analytical data through B2B partnerships with AI video model companies.

Core Problem: As AI video generation becomes indistinguishable from reality, society needs scalable human training data to improve detection models. Current solutions lack engagement.

Target Customer:

  • Primary: Gen Z/millennial gamers (18-35) seeking entertainment + income
  • Secondary: AI model companies needing real-time training data

Market Impact: Creates a self-reinforcing loop - the more users play, the better AI video models become, the more valuable the data is to sponsors.

Main Features

  1. Swipe-to-Classify Interface – Intuitive Tinder-like UX for rapid video classification
  2. Dynamic Leaderboard – Real-time ranking by accuracy score
  3. Dual Monetization:
  • B2B Data Sales – Sell anonymized user classification patterns + confidence metrics to video generation companies
  1. Reward System – Points → redeemable rewards funded by AI model sponsors
  2. Analytics Dashboard – Track personal accuracy rates, time-per-classification, confidence levels, and earned rewards

User flow:

  1. User lands on app → sees leaderboard
  2. User swipes through 5-10 videos (real/AI mix)
  3. Gets instant feedback on correctness
  4. Climbs leaderboard in real-time
  5. Can invite for referral rewards
  6. Backend aggregates: classification patterns, confidence metrics, demographic trends, and video-specific difficulty data flowing to sponsor dashboard

Business Model:

B2B Data Sales: AI companies sponsor rewards in exchange for anonymized classification data (accuracy rates, time-to-decision, video difficulty metrics, confidence distributions).

User Engagement Loop: Users earn points for accurate/fast classifications, share achievements on social media, bringing new players organically with near-zero CAC.

Scaling: As a one-person operation, the platform scales through user-generated engagement with no moderation overhead. More users = more valuable data = higher sponsor deals.

None, we started from scratch

AI Tinkerers Figma Google Lovable Station F

live demo

Summarizing URL...