Hacker News Reader

However - Automation bias is a common problem (predating AI), the 'human-in-the-loop' ends up implicitly trusting the automated system.

At least pre-LLM automation was written by a careful human who's job was on the line, and was deterministic.