The U.S. Department of Health and Human Services (HHS) just sent a clear signal that it wants to accelerate AI in medicine: it announced a Request for Information (RFI) focused on how AI could reduce healthcare costs and drive adoption in clinical care. In plain terms, HHS is asking the market—hospitals, insurers, researchers, vendors, patient advocates—what works, what doesn’t, and what guardrails are needed before AI becomes routine.
RFIs aren’t laws, but they’re often the first move in a much bigger chess game: defining priorities, mapping risks, and shaping the eventual rules and funding pathways.
Why “cost” is the headline
Healthcare AI discussions often drift toward futuristic diagnostics. But HHS putting cost reduction front and center suggests the government is targeting the messy, expensive reality of care delivery—where inefficiency is everywhere:
- administrative burden and documentation
- prior authorization and billing friction
- missed follow-ups and poor care coordination
- avoidable ER visits and readmissions
- staff burnout and shortages
If AI can reliably chip away at even a few of these, the savings could be meaningful. The key word, though, is reliably—because healthcare is full of tools that look great in pilots and disappoint at scale.
What “adoption in clinical care” really means
This isn’t about flashy demos. Adoption means AI has to survive the real-world gauntlet:
- Integration into existing EHR workflows (where good ideas go to die)
- Trust from clinicians who are already overloaded
- Clear accountability when AI is wrong
- Evidence that it improves outcomes, not just throughput
- Equity and bias controls so benefits don’t cluster in already-advantaged populations
- Privacy and security strong enough for sensitive patient data
If the RFI is framed correctly, it can surface where the friction actually is—and what incentives might remove it.
The biggest tension: automation vs. medicine’s “last mile”
AI is excellent at pattern recognition and drafting text. Medicine is full of edge cases, context, and human factors—patients who don’t fit the dataset, symptoms that don’t follow the textbook, and decisions that are as much about values as they are about probabilities.
So the near-term sweet spot may not be “AI replaces clinicians.” It’s more likely:
- AI reduces clerical work (notes, summaries, coding support)
- AI improves decision support (risk flags, guideline reminders)
- AI helps manage population health (outreach prioritization, care gaps)
- AI smooths operations (scheduling, capacity planning)
The win isn’t magic diagnosis. It’s fewer bottlenecks.
What to watch next
If HHS follows the RFI with concrete next steps, the impacts could show up in how healthcare organizations buy and deploy AI. Look for moves around:
- evaluation standards and validation requirements
- procurement guidance for federal and federally-funded programs
- policy positions on transparency, explainability, and auditability
- incentives tied to measurable cost and quality outcomes
Bottom line
HHS launching an RFI on AI in healthcare is a strong indicator that the conversation is shifting from “Should we use AI?” to “How do we use it responsibly, at scale, and with real savings?” The hard part won’t be inventing algorithms. It’ll be proving value in messy clinical reality—and building trust, governance, and workflows that make AI feel like a tool clinicians can actually live with.
