1 articles with this tag
Yoko Li's TetrisBench project reveals how LLMs, initially struggling with direct play, develop surprising, distinct strategies when tasked with generating game logic, outperforming most humans but faltering against top players' adaptive chaos.
