Hong-Zhan Lin

Tool-using agents frequently act without sufficient evidence, even when capable of strong static assessment, due to premature execution and incomplete investigation.

This work examines what structure is supplied by the game or workflow, what AI learns or produces, which capabilities and artifacts transfer across settings and roles, and what evidence supports the claims and identifies cross-role connections.

Speech QA performance peaks at 4.17 Hz, challenging the assumption that higher frame rates always yield better reasoning outcomes.