研究议程Research agenda

应用人工智能研究Applied AI research

APEX围绕智能体、风险判断与教育社会应用组织长期研究。每个方向通过具体计划、实验系统和公开成果持续更新。APEX organizes continuing work across agentic systems, risk judgement and applied AI in education and society. Each area develops through initiatives, experimental systems and public outputs.

长期研究方向Long-term research areas

方向描述长期问题；项目、系统和论文记录这些问题目前推进到了哪里。Areas define continuing questions; initiatives, systems and papers record current progress.

R—01

智能体与
自主系统Agentic AI
Systems

研究智能体如何理解任务、调用工具、协同工作并接受评估，以及人在何处参与和接管。How agents understand tasks, use tools, collaborate and undergo evaluation—and where people participate or take control.

当前问题Questions: 任务边界、工具调用、Multi-Agent 协作、人机复核Task boundaries, tool use, multi-agent collaboration and human review
实验系统System: APEX Stats Agent
当前状态Status: 研究流程原型与行为评估Research workflow prototype and behavior evaluation

查看 APEX 统计 AgentExplore APEX Stats Agent↗

R—02

决策与
风险智能Decision &
Risk Intelligence

研究复杂系统中的依赖关系、时间变化与方向性信号，以及模型判断如何形成可审查的解释路径。Dependency relationships, temporal change and directional signals in complex systems, with reviewable paths from model output to judgement.

当前问题Questions: 依赖结构、方向与时间、不确定性、解释与审查Dependencies, direction and time, uncertainty, explanation and review
重点计划Initiative: AEGIS
当前状态Status: 方法框架与研究原型方向Method framework and research prototype direction

进入 AEGIS 风险智能Explore AEGIS Risk Intelligence↗

R—03

教育、商业与
社会智能AI for Education,
Business & Society

研究人工智能如何进入教育选择、人才发展、计算商业和社会影响场景，并保留证据、人工复核与责任边界。How AI enters education choices, talent development, computational business and social contexts while preserving evidence, human review and responsibility.

当前问题Questions: 路径条件、证据时效、个性化判断、人工复核Pathway conditions, evidence freshness, personalized judgement and human review
应用研究Initiative: 教育路径智能Pathway Intelligence
既有记录Record: AI留学智能评估系统1.0AI Education Assessment System 1.0