AIへの依存が、社員の思考力を奪っている？

This article explains the risk of 'AI-based telephone game' effects—where repeated use of generative models and training on synthetic outputs causes divergence from original facts (model collapse)—and recommends a fundamental shift in design and usage policies: limit employee AI use to high-value tasks and rely on structured, fact-checked processes (such as specific, verifiable questions in hiring) to prevent knowledge decay.