CAPTCHAs are designed to differentiate humans from bots — but what if we flipped that idea on its head? This project proposes a “Reverse CAPTCHA,” where the system generates outputs and challenges a human to prove whether the source is artificial or not.
This playful and clever twist introduces students to natural language generation (NLG), text synthesis, and UX deception design. The system could generate short texts, images, or even voice clips, and then ask users: “Was this created by a human or an algorithm?”
You could use GPT-powered APIs (like OpenAI or Hugging Face) to generate outputs, mix them with real human-written samples, and display them through a gamified web interface. Users would then guess whether the input was bot-generated or authentic. Scores, leaderboards, and insights into detection accuracy could make this a great student engagement platform.
You’ll explore AI model integration, user experience design, and probabilistic testing methods — all within a humorous, creative framework that gets people questioning the limits of human-like computing.
- Explore Hugging Face Text Generation
- GPT-3 intro from OpenAI: OpenAI Text API
- Learn about CAPTCHAs: How CAPTCHA Works
Summary
- Title: The Reverse CAPTCHA – Can You Fool a Human Checker?
- Technology Stack: Python (Flask/Django), GPT APIs, Web Frontend (HTML/CSS/JS)
- Preferred Team Size: 2–3 students
- Categories: AI, NLP, Web Apps
- Tags: CAPTCHA, GPT, Language Generation, UX, Human-AI Interaction