SCICONVBENCH: Benchmarking LLMs on Multi-Turn Clarification for Task Formulation in Computational Science

A benchmark for evaluating how well LLMs ask clarifying questions when formulating computational science tasks, relevant to improving AI reliability in scientific workflows.

Auto-scored: A research benchmark that modestly improves understanding of LLM clarification behavior, with no direct risk or major mitigation.