人机交互：2025–2026 年度综述

An annual survey of the cs.HC arXiv category, covering research published from September 2025 – August 2026. Built on a close-reading set of 120 papers and organised into 44 chapters, it maps the year by the field's research sub-directions.

Book framework (share of the year's output):

• LLM-Mediated Interaction Design and Human-AI Teaming (78.9%) — This subfield investigates how large language models and agentic systems reshape interaction design, collaboration, and user support across diverse applications.
• Multimodal Data Visualization and Temporal Signal Interpretation (8.7%) — This subfield develops techniques for integrating multimodal, time-varying data streams into visual analytics and interpretable AI systems.
• Generative AI in Education and Learning Analytics (6.4%) — This subfield explores the design, deployment, and evaluation of generative AI tools for personalized learning, tutoring, and educational practice.
• Social Media Dynamics and Algorithmic Influence (3.3%) — This subfield examines how algorithmic curation and AI-mediated content shape user engagement, perception, and behavior on social platforms.
• Trustworthy and Human-Centered Health AI (0.9%) — This subfield focuses on developing and assessing AI systems for healthcare that are trustworthy, safe, and responsive to clinical and patient needs.
• Real-Time Feedback and Embodied Training Systems (0.5%) — This subfield designs and evaluates interactive systems that provide in-situ, real-time feedback for skill acquisition in physical and professional training.
• Human-Robot Collaboration and Shared Mental Models (0.5%) — This subfield studies how humans and robots/agents coordinate, build shared understanding, and collaborate effectively in team settings.
• Autonomous Driving and Agentic Vehicle Interaction (0.4%) — This subfield addresses driver-vehicle interaction, trust, and control in automated and autonomous driving systems.
• Behavioral Sensing and Cognitive State Intervention (0.3%) — This subfield develops unobtrusive sensing and intervention methods to detect and influence cognitive and behavioral states in daily life.
• LLM-Based Decision Support and Creative Prototyping (0.2%) — This subfield explores how LLM-powered chatbots and multi-agent architectures support decision-making, creative processes, and rapid prototyping.

Each sub-direction is covered across three chapters — how it framed its questions, the strength of its key findings, and its open debates and frontiers — so the reader gets a traceable picture of what this part of the field actually produced over the year.