The inquiry should begin with the decision leadership needs to make, not with the availability of an automated call. Questions, permitted sources, scoring rules, exceptions, and ownership of the response should be defined before outreach begins.
Begin with the management decision
A location inquiry is useful only when it is connected to a decision someone is authorized to make. The design record should identify the intended decision, the selected locations, the time window, and the person responsible for reviewing the evidence.
A question that cannot change a training, resource, service, or follow-up decision may still be interesting, but it should not be presented as operational intelligence.
Specify the inquiry before contact
The same core questions should be used across the approved scope, with controlled branching for clarification. The record should distinguish the words reported by a location from classifications or scores added later by a system or reviewer.
- Purpose and intended decision
- Selected locations, timing, language, and permitted contact path
- Question wording and allowed clarifications
- Source record, missing-response handling, and escalation rules
- Scoring rubric, version, reviewer, and exception process
Keep facts, claims, and inference separate
A reported answer is not automatically a verified fact. A reviewable record should identify whether a finding is directly observed, reported by a participant, derived through a stated rule, or inferred by a model. Missing and conflicting information should remain visible rather than being silently resolved.
A score can help organize attention, but it should never conceal the response, the rule that produced the score, or the uncertainty a reviewer needs to understand.
Evaluate the method before expanding it
A limited evaluation can examine response coverage, completeness, consistency across reviewers, exception frequency, and whether the information helped an authorized person make the intended decision. These measures assess the inquiry process; they do not establish a customer result or prove that an operational intervention worked.
Limits of the current method
Location responses may be incomplete, self-reported, affected by local conditions, or shaped by question wording. Language handling, transcription, sampling, and scoring can introduce additional error. A structured inquiry is not an audit, a substitute for direct observation, or proof that every location has been represented fairly.
This note documents a working method under development. American Intelligence Labs is not reporting a deployed customer program, validated result, or completed field study.
Sources and reference frameworks
- Artificial Intelligence Risk Management Framework (AI RMF 1.0)
National Institute of Standards and Technology. A voluntary framework for incorporating trustworthiness considerations into the design, development, use, and evaluation of AI systems.
- Artificial Intelligence: An Accountability Framework for Federal Agencies and Other Entities
U.S. Government Accountability Office. An accountability framework organized around governance, data, performance, and monitoring, with review questions for system owners and assessors.