D.5 Human Judgment

Where is human attention going, and what do we keep deciding again?

Human attention is scarce, but some work should remain human. Novel architecture, ambiguous evidence, and consequential review may require human judgment. Look instead for knowledge people repeatedly reconstruct or decisions they repeatedly make that could economically become durable.

Ask two questions:

Where is attention being spent? Distinguish creation of new intent, evidence interpretation, reconstruction of known context, routine review, and exception handling.
Which judgments recur in substantially the same form? Ask whether a model, alignment mechanism, procedure, or other durable structure could keep people from making the same decision again.

  HUMAN JUDGMENT                          "Where is scarce attention going?"
  ───────────────────────────────────────────────────────────────────────────
  Novel intent / architecture     expected human work
  Semantic validation             often intentional
  Repeated reconstruction         investigate externalization
  Routine adjudication            investigate enforcement
  Recurring emergency             diagnose the class
  ───────────────────────────────────────────────────────────────────────────
  Repeated ≠ automate automatically.
  Ask whether making it durable will repay its carrying cost.

The goal is less avoidable repeated reasoning per unit of durable output, while people remain responsible for work whose novelty, ambiguity, consequence, or poor governability still requires judgment.