My research group:

Overarching Research Themes

*Extracted by GPT-4, there may be inconsistencies.* #### *Ethics and Human-Centric AI* My research group explores the ethical implications and responsibilities of AI systems in our society. We investigate the biases and social implications of AI technologies, as exemplified in the paper [Disparities in LLM Reasoning Accuracy and Explanations: A Case Study on African American English](https://arxiv.org/abs/2503.04099), which reveals significant bias in large language models. Another notable contribution is the paper [Mind the Gesture: Evaluating AI Sensitivity to Culturally Offensive Non-Verbal Gestures](https://arxiv.org/abs/2502.17710), addressing how AI can misinterpret or perpetuate biases through non-verbal cues. Our ongoing research includes frameworks for understanding users' perceptions of AI, as demonstrated in [User-Driven Value Alignment: Understanding Users' Perceptions and Strategies for Addressing Biased and Discriminatory Statements in AI Companions](https://arxiv.org/abs/2409.00862). #### *Narrative Structures and Empathy* My research group explores the intricacies of narratives and their emotional impacts, particularly in online contexts. We delve into how narrative styles can evoke empathy, shown in the paper [HEART-felt Narratives: Tracing Empathy and Narrative Style in Personal Stories with LLMs](https://arxiv.org/abs/2405.17633), which investigates how personal storytelling resonates with different audiences. Another key publication, [Quantifying the narrative flow of imagined versus autobiographical stories](https://www.pnas.org/doi/10.1073/pnas.2211715119), sheds light on the structural elements that differentiate types of narratives. Additionally, we explore users’ engagement with stories in various online spaces, as highlighted in [Where Do People Tell Stories Online? Story Detection Across Online Communities](http://arxiv.org/abs/2311.09679). #### *Social Intelligence in AI* My research group explores the development of socially intelligent AI agents, focusing on how these systems can better understand and interact within social contexts. Our recent work includes the paper [SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents](https://arxiv.org/abs/2310.11667), which presents a framework for assessing social reasoning in AI interactions. Furthermore, we investigate the nuances of language usage in social interactions through [Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models](https://arxiv.org/abs/2305.14763), which tests the limitations of AI in simulating social reasoning. Our research also includes practical implementations, as seen in [ProsocialDialog: A Prosocial Backbone for Conversational Agents](https://arxiv.org/abs/2205.12688), aimed at fostering healthier interactions between users and AI systems. #### *Moderation and Toxic Language Detection* My research group explores new methodologies for detecting and moderating toxic language online, recognizing its critical importance for community safety. A major aspect of our work is highlighted in the paper [PolyGuard: A Multilingual Safety Moderation Tool for 17 Languages](https://arxiv.org/abs/2504.04377), which seeks to create a more inclusive approach to handling toxic language across diverse linguistic contexts. Additionally, we analyze how biases affect detection systems, as detailed in the paper [BiasX: 'Thinking Slow' in Toxic Language Annotation with Explanations of Implied Social Biases](https://arxiv.org/abs/2305.13589), addressing how societal biases can distort the interpretation of language. Our research framework also encompasses contextual analysis of offensive statements, evident in the study [COBRA Frames: Contextual Reasoning about Effects and Harms of Offensive Statements](http://arxiv.org/abs/2306.01985).