SOTOPIA-$π$: Interactive Learning of Socially Intelligent Language Agents
This work addresses the gap in social learning for language agents, offering a method to enhance their social skills, though it is incremental as it builds on existing techniques like behavior cloning and reinforcement learning.
The paper tackled the problem of improving social intelligence in language agents by proposing SOTOPIA-π, an interactive learning method using behavior cloning and self-reinforcement training on filtered social interaction data, which enabled a 7B LLM to match the social goal completion ability of a GPT-4-based agent while enhancing safety and maintaining general QA performance.
Humans learn social skills through both imitation and social interaction. This social learning process is largely understudied by existing research on building language agents. Motivated by this gap, we propose an interactive learning method, SOTOPIA-$π$, improving the social intelligence of language agents. This method leverages behavior cloning and self-reinforcement training on filtered social interaction data according to large language model (LLM) ratings. We show that our training method allows a 7B LLM to reach the social goal completion ability of an expert model (GPT-4-based agent), while improving the safety of language agents and maintaining general QA ability on the MMLU benchmark. We also find that this training paradigm uncovers some difficulties in LLM-based evaluation of social intelligence: LLM-based evaluators overestimate the abilities of the language agents trained specifically for social interaction.