Why wording changes the evidence needed
“Helps Quality” is too broad. State what the assistant does, with which information, for which users, before which decision, and with which limits. Context includes the user’s expertise, workload, available sources, and consequences of an incorrect answer.
Worked example
Worked example — fictional
Broad: “AI validates protocols.”
Bounded: “The assistant suggests candidate inconsistencies between a draft protocol and an approved requirement set for review by qualified validation personnel. Reviewers independently verify source support and retain responsibility for the protocol review. The assistant does not approve documents, establish requirement completeness, or determine production release.”
The bounded statement makes evaluation possible. Test candidate findings, source support, unsupported cases, and the reviewer workflow. Do not infer suitability for a broader task from that evidence.
Compare two contexts
An experienced reviewer using the assistant after completing an independent review may rely on it differently from a novice who treats the generated list as a complete checklist. The same model output can lead to different human decisions.
Exercise
The pilot succeeds with five expert reviewers. Can the team release it to new graduates without further consideration?
Show suggested answer
Evaluate the changed users and reliance. Training, interface guidance, source access, task design, and observed reviewer performance may require additional evidence. A model score does not demonstrate equal workflow performance across user groups.
Expected records
An intended-use statement linked to users, source inputs, outputs, reliance, exclusions and approval responsibility. Adapt record names and detail to the applicable procedures; these are practitioner examples, not a universal mandatory document list.
Relevant regulations and guidance
Regulations set requirements within their scope; guidance offers recommendations or explains applicable expectations; standards depend on adoption or incorporation. Voluntary frameworks do not establish compliance. The lesson's controls, test designs and example records are practitioner suggestions unless explicitly identified otherwise.
Guidance · U.S. Food and Drug Administration · February 2026
Scope: United States · Medical-device manufacturers assuring software used in production or their quality management system.
Relevance to this lesson: Relates the assurance approach to the intended use of in-scope production/QMS software.
Publication status: Final / published. Source checked 21 September 2026.
Publication identity, scope and status checked against the linked official source. This is not an independent audit of every retained guide statement.
Source details and applicabilityVoluntary framework · National Institute of Standards and Technology · NIST AI 100-1 · January 2023
Scope: United States · Cross-sector AI governance and risk management, including generative AI.
Relevance to this lesson: Supports defining context before selecting risk measures.
Publication status: Final / published. Source checked 21 September 2026.
Publication identity, scope and status checked against the linked official source. This is not an independent audit of every retained guide statement.
Source details and applicability