Bingsheng Arthur Yao

I am an associate research scientist in the Khoury College of Computer Sciences at Northeastern University (PI: Prof. Dakuo Wang). I received my PhD from Rensselaer Polytechnic Institute, advised by Prof. Jim Hendler.

I study how LLM agents can genuinely collaborate with humans through the lens of remote human collaboration. My research sits at the intersection of Human-Computer Interaction and Natural Language Processing, with a focus on designing the interaction patterns and methodologies for effective human-agent collaboration in practice, and on developing and evaluating LLM agents that can think and behave collaboratively.

Research

I. Design and Study Human-Centered AI Systems for Clinical Care

Healthcare workflows often break down at the seams between provider teams, patients, and caregivers. I design, deploy, and study AI/LLM-powered multi-modal systems that work in those gaps to support diverse stakeholders involved in different clinical settings. My work spans clinical decision-making, patient-provider communication, provider coordination, remote patient monitoring, and post-surgical care.

• RECOVER (CSCW '26) — LLM-based remote patient monitoring for post-surgical GI cancer care
• Serious Illness Conversations (CHI '26) — Balancing efficiency and empathy in serious-illness conversations in the ED
• Collaboration Breakdown (CHI '26) — Within provider teams and between patients in post-surgery care
• CardioAI (CHI '25) — Multimodal AI plus wearable for cancer treatment-induced cardiotoxicity
• Talk2Care (IMWUT '24) — Asynchronous communication between older-adults and care providers
• Sepsis Diagnosis Support (CHI '24) — AI-assisted clinical decision making for sepsis diagnosis

II. Genuine Human-Agent Collaboration

I envision a near future where LLM agents can genuinely work with us, like a remote human collaborator. To get there, I design interaction patterns and study methodologies for human-agent collaboration, and develop and benchmark LLM agents that can think and behave collaboratively. My work spans collaboration frameworks and design philosophy, trust and shared context in human-agent teams, evaluation methods for collaboration quality, agent oversight, and role-playing agents.

• Configurable Research Platform for Human-Agent Collaboration (CHI '26)
• Dark Patterns Meet GUI Agents (CHI '26) — Examining LLM agent susceptibility to manipulative interfaces
• CHI '26 Workshop "Human-Agent Collaboration" (CHI EA '26) — A vision, design philosophy, and empirical framework for treating LLM agents as remote human collaborators
• Agent A/B (CHI EA '26) — Automated and scalable A/B testing on live websites with interactive LLM agents
• Dynamic Persona Refinement Framework (preprint, 2025) — Iteratively optimizing behavioral and cognitive alignment
• Survey of LLM Role-Playing Agent Evaluation (ACL Findings '25) — Guidelines for evaluating LLM role-playing agents

Note

Please refer to my Google Scholar page for the most up-to-date publication record.

The best way to reach out is through emails: b [dot] yao [at] northeastern [dot] edu.

---