RC's Blog
The Verification Tax
The Verification Tax argues that AI make teams faster while eroding the expertise those teams need to catch what AI get wrong. The core problem is that verification work is invisible, a checked answer and a lucky unchecked one look identical in the final output, so this erosion is the outcome unless teams design against it. The fix requires two things together: training people's judgment through deliberate friction, and building real checks into the work itself (source grounding, refusal thresholds, mandatory review before anything ships) rather than relying on individual attention, which is exactly what erodes over time. By framing The Verification Tax as a decision every team is already making, the question stops being whether to use AI and becomes whether you've built the checks to catch what it gets wrong.