About Typed
What is Typed?
Typed is a mix between personality tests like MBTI and BuzzFeed quizzes, and interactive data experiences like Spotify Wrapped. I made it because most personality tests feel too long, repetitive, and predictable.
Instead of asking direct personality questions, Typed experiments with a new question: can your media preferences say something about your personality? The experience uses music, movies, shows, artists, actors, and head-to-head choices to build a profile based on what you actually like. The hypothesis is people who have very similar interests in media will have similar personality traits.
The Problem
Traditional tests can feel boring because they ask obvious questions and repeat the same patterns. They also may not understand what you like.
The Experiment
Typed turns the test into a game by using taste-based choices on movies, TV shows, and musicinstead of standard personality prompts.
The Result
An AI model uses your choices to categorize your type and generate a personalized result based on your actual preferences and behavior.
▶Conceptual and Technical Statement
Typed is an interactive personality experiment that combines media preferences, competitive selection mechanics, and AI-generated analysis into a single experience. The project was inspired by traditional personality tests such as MBTI and BuzzFeed-style quizzes, as well as data-driven experiences like Spotify Wrapped. While those formats remain popular, many personality tests have become repetitive and predictable because they rely on direct questions that make their intentions obvious. I wanted to explore whether personality could instead be inferred indirectly through taste and decision-making.
Rather than asking users to describe themselves, Typed analyzes the music, movies, shows, artists, and actors they choose throughout the experience. The project is structured more like a game than a survey, using tournament brackets, “keep/bench/cut” systems, and head-to-head comparisons to make interaction more engaging and less repetitive. This approach was designed to make the process entertaining while still collecting meaningful preference data.
On the technical side, the project was built using Next.js, React, and TypeScript. Deezer, Last.fm, and the iTunes Search API were used to dynamically generate music selections based on user choices. TMDB was used for movie and TV show data. Framer Motion was used to create animated transitions and drag-and-drop interactions between stages, helping the experience feel fluid and interactive. User selections are stored client-side throughout the session via localStorage, then sent to GPT-4o, which assigns the user to one of 16 personality archetypes and generates a personalized result based on both broader patterns and specific choices. The project ultimately explores how algorithms can construct identity through entertainment and media consumption.