SciGuard is proposed, an agent-based guardrail that employs large language models, tools and external knowledge to assess and control risks in scientific AI interactions, and achieves a state-of-the-art harmlessness score on red-teaming queries, while maintaining high performance on benign tasks, without sacrificing sc...
Jiyan He, Hao-Xiang Guan, Weitao Feng et al.· AI for Science· 26 citations· ⚡4
A deeper understanding of AI assistance in education is provided by showing how differences in learner profiles and learning behavior shape who benefits from AI-supported learning, and how learner profiles and learning behavior shape these patterns.
Jing-Wei Yi, Yueqi Xie, Jiyan He et al.· arXiv.org· 0 citations
OrchBench is established as an efficient and interpretable benchmark for comparing and diagnosing multi-agent orchestration plans, finding that preserving task-critical information is more important than simply increasing the number of agents, and the benefits of parallelism diminish as coordination failures accumulate...