Synthetic Deep Web: On Algorithmic Invisibility in the Age of Generative Models

The author argues that generative language models are creating a "synthetic deep web" or Deep Web 2.0, where content is rendered invisible not by technical indexing failures, but by the models' safety alignments, training corpus distributions, and anti-advertising reflexes. Unlike the classical deep web, this new layer of algorithmic invisibility is semantic and deliberate, leaving users with the illusion of complete and exhaustive answers while filtering out entire categories of commercial, political, and specialized information. Open article →