School of Engineering Department of Computer Science and Engineering 129 Designing Conversational Agents for Neurocognitive Disorders Screening Supervisor: MA Xiaojuan / CSE Student: ZHANG Li / COMP Course: UROP 1100, Spring Conversational agents (CAs) possess significant potential to automate the early screening of neurocognitive disorders (NCDs) and enhance accessibility to these services, enabling timely and critical intervention. This study investigates four prompting strategies, ranging from structured chain-of thought (CoT) scaffolding to minimal instructions across two large language models—GPT-4o and Qwen, across ideal, noisy, and realistic automatic speech recognition (ASR) conditions. Results showed highest appropriate response rates (>90%) with most structured prompts. Performance decline was observed with noisy data and with less structured prompts, with Qwen demonstrating higher robustness to ASR errors. These findings highlight the effectiveness of CoT and scaffolding strategies and the importance of enhancing model resilience to noisy input for reliable, real-world CA-based NCD screening. Interaction Design for Human-AI Collaboration Supervisor: MA Xiaojuan / CSE Student: ZHAO Yuhua / COMP Course: UROP 1100, Summer This study examines the implementation of artificial intelligence (AI)–based psychological counselors utilizing the empty chair technique to mediate and resolve interpersonal conflicts among college student couples. Grounded in the principles of Gestalt therapy, the empty chair method enables individuals to engage in perspective‐switching dialogues, fostering emotional expression, self‐reflection, and empathy. The proposed AI system integrates advanced natural language processing, multimodal emotion recognition, and adaptive conversational frameworks to guide participants through structured therapeutic interactions. A series of intervention sessions were conducted with college couples experiencing recurrent disputes, during which the AI counselor facilitated the alternation of roles and the articulation of underlying emotional needs. Data collection included both qualitative feedback and pre‐ and post‐intervention assessments of communication quality and relationship satisfaction. Initial findings suggest that AI‐mediated application of the empty chair technique can effectively reduce emotional volatility, promote mutual understanding, and support constructive conflict resolution. These results contribute to the emerging field of AI‐assisted psychotherapy and provide a foundation for the development of culturally responsive, scalable counseling solutions tailored to young adult romantic relationships.
RkJQdWJsaXNoZXIy NDk5Njg=