Detecting Soft Errors in Parallel Software with LLM-tuned Instruction Duplication

A research paper proposes using LLMs to tune instruction duplication for detecting soft errors in parallel software, a reliability engineering technique with no direct AI risk implications.

Auto-scored: This is a technical reliability paper with no harmful AI capability, misuse, or governance implications.