More than a Judge: Empirical Study of Agent-Human Interaction in Crowdsourced Testing

arXiv·medium signal

In crowdsourced testing, the volume and uneven quality of submitted bug reports create a heavy review burden. This empirical study examines agentic AI not just as an autonomous 'judge' of report quality but as a collaborator with human reviewers, characterizing where the human-agent handoff actually helps. Useful framing for builders designing human-in-the-loop triage or review systems. Published 2026-06-04 (cs.SE).