Hospitals and insurers are already using AI to triage, price, and decide. MB is asking them to show their work.
What Happened
Hospitals and insurers now use AI across documentation, triage, risk prediction, scheduling, billing, prior authorization, and clinical decision support. Federal Section 1557 nondiscrimination rules already reach patient-care decision-support tools, and current health-policy research warns that biased data, design, and deployment can let these systems reproduce racial and ethnic disparities rather than remove them. Advocacy groups, including the NAACP, have begun pushing for equity-first AI standards in medicine.
The Standing Questionnaire
MB is putting the same ten questions, in writing, to every hospital, insurer, AI vendor, and digital-health company we cover:
What decision does the tool influence?
Is the tool clinical, administrative, financial, or all three?
Which patients were represented in training and validation data?
Was it tested on this hospital's local patient population before use?
Are results measured by race, ethnicity, language, disability, age, sex, gender identity, payer, and ZIP code where relevant?
Can a clinician override the output — and is that override easy to document?
Is the patient told AI is being used?
Who investigates harmful recommendations or missed care?
How often is the model changed, retrained, or revalidated?
Will the institution publish an equity-impact report?
The Tracker
Every response gets logged publicly as: Disclosed · Partially disclosed · Declined to answer · No public policy found · Evidence of local equity auditing. The goal isn't a single “AI may be biased” story it's a repeatable, named standard applied institution by institution.
Sources
[1] The Growing Use of Artificial Intelligence in Health Care and Implications for Disparities — KFF
[2] The future of algorithmic nondiscrimination compliance — PMC
[3] NAACP pressing for 'equity-first' AI standards in medicine — Reuters