Human Taste: Content

Evaluate writing and content through human taste -- the trained judgment that detects whether text is clear, authentic, appropriately dense, and tuned to its audience.

This complements the UX-focused human-taste and code-focused human-taste-code skills. For full citations see references/research-sources.md.

Why This Matters

AI can generate fluent text, but fluency is not quality. Studies show AI text detectors achieve only ~58% accuracy -- meaning AI-generated content often passes surface inspection. The real differentiator is taste: the ability to detect when writing is generic, over-hedged, structurally predictable, or tonally flat, even when it is grammatically perfect.

Key insight: AI slop is not a vocabulary problem. It is a thought-structure problem -- predictable arcs, tidy conclusions, and the absence of genuine voice.

Quick Start

When asked to evaluate content: