Substack article by Daniel Stecher
Jason Bloomberg published a piece declaring that Human in the Loop is fundamentally broken. His argument is clean: automation bias, rubber stamping, governance lag, accountability laundering. Organizations build oversight for the systems they wish they had, not the ones they actually run. He proposes inverting the relationship entirely. Stop placing humans as checkpoints inside automated flows. Treat automation as an instrument inside human workflows.
He is right about the diagnosis. He is writing from the wrong floor of the building.
Bloomberg is describing agentic AI governance in enterprise software. What he cannot see from where he is standing is the operations floor at 03:00, where the most vulnerable moment in any airline’s day is staffed by whoever is available, where the systems making crew decisions run on architecture designed the same decade as the people who just retired, and where the gap between what the controller knows and what the organization permits is exactly where the cascade begins.
That gap has a name in my work. The Gap Between Recognition and Permission. It is not a technology problem. It is not a psychology problem. It is a structural condition that no governance framework, however elegantly theorized, has yet resolved in live operations.
Click here to read the entire article.
