Bench Market

Benchmarket supports a transparent interactive ecosystem for online gaming versus state of the art AI models. For the purposes of this hackathon I enabled every model available on Groq to be utilized and to have its performance measured in the classic board game Battleship. A player can select one of the models available on the menu and play against it. As the game plays out, the AI model can offer odds to the opponent to make the game more interesting. The result of the matchup is recorded and stands as a testament to the effectiveness of the AI to both play the game and judge its likelihood of winning throughout the game.

Benchmarket is built on the cutting edge of blockchain technology. The stack combines Web3Auth with the Fhenix and Galadriel Blockchains, making use of Multi-Party Computation (MPC), Account Abstraction, Fully Homomorphic Encryption (FHE), Artificial Intelligence (AI), and Trusted Execution Environment Machine Learning (TEEML)

Web3 auth provides a seamless gaming experience for use of the Fhenix blockchain. The player does not have to confirm every transaction while playing the game, making for a better UX.

The Fhenix blockchain protects private data from prying eyes, which in this case means that a player in the Battleship game can only see their own battleships and cannot see their opponents battleships. The concept is familiarly called "Fog of War" in gaming.

I have written a script that can be used with Node.JS to connect the two blockchains. Upon someone creating a game on Fhenix, the script privately places ships on Fhenix, then once a player makes a move the script makes calls to Galadriel to perform AI actions.