Semantic ablation: why AI writing is generic and boring, explained

A Register opinion piece introduces "semantic ablation" as a term for how AI systematically strips high-entropy information from text during refinement. The process replaces unique metaphors and precise language with generic, high-probability token sequences across three stages: metaphoric cleansing, lexical flattening, and structural homogenization. The author argues vocabulary diversity collapses through successive AI refinement loops, performing a "systematic lobotomy" on original writing. The concept provides a technical framework for why AI "polishing" often destroys the distinctive qualities that made writing effective.

View full digest for February 18, 2026