The article investigates the potential of Large Language Models (LLMs) to exhibit Theory of Mind (ToM) abilities, particularly in Human-Robot Interaction. ToM involves attributing mental states such as beliefs, intentions, desires, and emotions to oneself and others, and is a vital component of human-like interaction. The authors focus on Perceived Behavior Recognition, where a robot employs an LLM to assess its generated behavior in a manner similar to a human observer. The study showed promising initial results; however, subsequent perturbation tests revealed limitations in the LLMs’ abilities.
Publication date: 12 Jan 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2401.05302
