[NLP Seminar] Artificial Social Intelligence? On the challenges of Socially Aware and Culturally Adaptable LLMs

Modern AI systems such as LLMs are pervasive and helpful, but do they really have the social intelligence to seamlessly and safely engage in interactions with humans? In this talk, I will delve into the limits of social intelligence of LLMs and how we can measure and anticipate their risks. First, I will introduce Sotopia, a new social simulation environment to evaluate the interaction abilities of LLMs as social AI agents. I will show how today's most powerful models struggle to socially interact due to inability to deal with information asymmetry. Then, I will shift to how LLMs pose new ethical challenges in their interactions with users. Specifically, through their language modality and possible expressions of uncertainty, we show that LLMs tend to express overconfidence in their answers even when incorrect, which users tend to over-rely on. Finally, I will introduce NormAd, a novel hierarchical framework to evaluate the cultural adaptability of LLMs, showing that while LLMs may have knowledge of cultures, they struggle to effectively leverage that knowledge in downstream reasoning. I will conclude with some thoughts on future directions towards socially aware and ethically informed AI.