A Demonstration of Blabrecs, an AI-Based Wordgame


  • Max Kreminski University of California, Santa Cruz
  • Isaac Karth University of California, Santa Cruz


Wordgames, Board Games, AI-based Game Design, Markov Chains, Neural Networks


Blabrecs is an AI-based modification to the popular wordgame Scrabble. In Blabrecs, English dictionary words may not be played; instead, players may only play nonsense words that are approved by a classifier trained on a list of English dictionary words. Gameplay therefore revolves around inventing plausibly English-sounding nonsense words and learning how to fool the classifier. In this paper, we briefly introduce our design goals for Blabrecs; describe the process by which Blabrecs was designed; and present two distinct implementations of the game's AI judge.




How to Cite

Kreminski, M., & Karth, I. (2021). A Demonstration of Blabrecs, an AI-Based Wordgame. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 17(1), 243-245. Retrieved from https://ojs.aaai.org/index.php/AIIDE/article/view/18917