EscapeBench: Towards Advancing Creative Intelligence of Language Model Agents
It addresses the problem of limited creative intelligence in language model agents for researchers, though it is incremental as it builds on existing agent frameworks.
The paper tackles the lack of benchmarks for creative adaptation in language model agents by introducing EscapeBench, a room escape game suite, and finds that current models achieve only 15% average progress without hints. It proposes EscapeAgent, which reduces steps and hints by up to 40% and executes over 1,000-step action chains.
Language model agents excel in long-session planning and reasoning, but existing benchmarks primarily focus on goal-oriented tasks with explicit objectives, neglecting creative adaptation in unfamiliar environments. To address this, we introduce EscapeBench, a benchmark suite of room escape game environments designed to challenge agents with creative reasoning, unconventional tool use, and iterative problem-solving to uncover implicit goals. Our results show that current LM models, despite employing working memory and Chain-of-Thought reasoning, achieve only 15% average progress without hints, highlighting their limitations in creativity. To bridge this gap, we propose EscapeAgent, a framework designed to enhance creative reasoning through Foresight (innovative tool use) and Reflection (identifying unsolved tasks). Experiments show that EscapeAgent can execute action chains over 1,000 steps while maintaining logical coherence. It navigates and completes games with up to 40% fewer steps and hints, performs robustly across difficulty levels, and achieves higher action success rates with more efficient and innovative puzzle-solving strategies.