Chun-Yan Zhu

Trustworthy Generation and Verification-Guided Correction for ChatGPT-Type Large Language Models: Symmetry-Aware Technical Mechanisms and Ethical Risk Analysis

Reliable retrieval-augmented generation requires consistency across query interpretation, evidence selection, and final answer generation. This study defines computational symmetry as bidirectional coverage among canonical query constraints, traceable evidence, and answer claims, with residual asymmetry triggering corr...