Description

The iNNk game is a web-based, multiplayer, drawing game where two or more people play together against a neural network (NN). To win the game, the players need to successfully communicate a secret code word to each other through drawings without being deciphered by the NN. Players are assigned one of the two roles during the game: the Sketcher and the Guesser. The Sketcher is tasked with drawing something based on the code word assigned by the game. The goal is to draw the code word so that the human Guesser can interpret the code word accurately before the NN. The Guessers are tasked with entering their guess of the code word based on the Sketcher’s drawing before the NN guesses correctly. The NN always plays the role of a Guesser, and its goal is to decipher correctly first.

Twitter: @iNNkGame

Try out the game on this page and see if your team can communicate safely without the neural network’s comprehension.

The game was developed in collaboration with user experience researchers from the Procedural Expression Lab at Drexel University.

Related Research

• iNNk: A Multi-Player Game to Deceive a Neural Network
• Dealing with Adversarial Player Strategies in the Neural Network Game iNNk through Ensemble Learning. Best Paper Award
• Player-AI Interaction: What Neural Network Games Reveal About AI as Play