Researchers have created a benchmark called TextQuests to evaluate the performance of Large Language Models (LLMs) in text-based video games. The benchmark consists of 20 different games, each with its own set of rules and objectives. According to the results, LLMs are able to achieve high scores in some games, but struggle in others due to limitations such as lack of common sense or inability to understand subtle language cues. The authors argue that this is because current LLMs are not designed to handle complex decision-making tasks. --- Why it matters: This matters to AI researchers and engineers because it highlights the limitations of current LLMs and the need for more sophisticated models that can handle complex decision-making tasks, which is crucial for applications such as game playing, dialogue systems, and other interactive tasks. Source: https://huggingface.co/blog/textquests