Description :

Yoon Kyung Lee, a postdoctoral researcher at out lab, visited the University of Southern California to deliver a presentation on how large language models (LLMs) have potential to improve human-AI interactions, with a focus on modeling cognitive processes to better align their responses with user expectations.

Her talk consisted of two sub-topics.
1. Modeling the Social-Cognitive Aspects of LLMs: She reviewed ongoing efforts focused at enhancing LLM reasoning capabilities, specifically on generating empathic responses and implementing targeted reappraisal in online peer support contexts. These advancements important, as they aim to make AI interactions more nuanced and human-like.
2. Exploring Real-World Applications and Limitations: She also examined the potential and challenges of using LLMs in real-world scenarios, such as health monitoring and social support systems. She discussed the potential of LLM application across diverse contexts that require a deep understanding of human intentions, emphasizing the need to improve their adaptability and reliability for everyday decision-making.
Her presentation calls attention to the transformative potential of LLMs in fostering more meaningful and effective human-AI collaborations.

---