Definition
The share of consequential operational decisions (pesticide applications, breeding and treatment actions, harvest timing, irrigation events, lot releases) whose record of decision carries the AI prescription, alert, or recommendation that contributed to it, with enough identity to reconstruct the contribution later.
The Measurement Gap
There is no metric that tells you what share of consequential decisions (application, breeding, treatment, harvest) carry a record of the AI contribution that shaped them, which means a drift complaint, an organic audit, or a claim cannot reconstruct what actually drove the decision.
Formula
ACRR = N_decisions_with_contribution_record / N_consequential_decisions, by decision type; a contribution record requires system identity, recommendation identity or value, and timestamp
Proposed formula, pending validation pilots.
At a Glance
Decision Supported
Whether a drift complaint, an organic audit, an insurance claim, or a food safety investigation could reconstruct what drove the decision, and which record types need the contribution field added.