Separating healthcare AI gold from coal: Insights from 3 industry leaders

CHIME and HLTH’s December “Drive to ViVE” conversation brought together three powerhouse CEOs to tackle a critical question: How do we ensure access to AI innovation across ALL care settings, especially those facing resource constraints?

Dr. Brian Anderson (Coalition for Health AI), Sachin Jain (Scan Health Plan), and Dr. Kyu Rhee (National Association of Community Health Centers) didn’t just throw around buzzwords on the promise of healthcare AI. They served up hard truths about separating vendors’ “fool’s gold” from real solutions that drive measurable value.

The Standards Perspective

Brian Anderson, Coalition for Health AI (CHAI), ViVE Speaker

When two major academic medical centers spent $5 million just to monitor a dozen AI models, they learned an expensive lesson: governance without the right structure is unsustainable.

Anderson’s key insight? Success requires cross-functional accountability. What they’re seeing now with multiple health systems is “creating a level of accountability with the service line or the business unit leaders. So, if it’s a clinical decision tool, it’s the cardiology chair or the surgery chair.” The principle: whoever owns the tool owns the performance metrics.

CHAI has developed AI model cards, nutrition labels for algorithms that reveal how models were trained, what populations they’ve been tested on, and where their limitations lie. For Anderson, these are now “table stakes” in vendor procurement.

But here’s where AI is already exceeding expectations: patients using tools themselves. Anderson points to “countless stories out there of patients putting in, you know, rare sun signs and symptoms and coming back with a diagnosis and lo and behold, Claude or ChatGPT or Gemini was … correct.” In some cases, lives have been saved. The challenge for providers? Partnering with patients in this new reality rather than resisting it.

MYTH: AI governance should live in the IT department
REALITY: The most successful models vest accountability with clinical and business unit leaders who own outcomes

The Implementation Reality

Sachin Jain, Scan Health Plan

“We’re in the first inning of a nine-inning journey,” Jain warns. Many are “checking the box on AI”, which is easy, he continued, but “fundamentally changing the work that people are doing and actually using these new capabilities to change the actual operations of an organization” is the hard part.

Scan Health Plan has ambient listening and decision support tools deployed across thousands of member service representatives. The technology works. But supervisors aren’t using new AI-enabled capabilities to monitor call patterns, build word clouds of member concerns, or spot systemic issues in real-time. Why? They haven’t been trained to think differently about their jobs.

Jain’s controversial move? Placing AI initiatives under Scan’s Chief People Officer, not the CIO. “All our learning and development efforts have been reoriented around AI because this is fundamentally about workforce transformation, not technology deployment.”

His vendor advice is blunt: Never sign contracts longer than one year. “The solution you bought in year one may be obsolete by year two. And there’s a nation of hucksters selling fool’s gold right now.”

For Jain, the real test isn’t whether organizations claim they’re “doing AI”—it’s whether patients say “my care is better” and workers say “my job is better.”

MYTH: We’re far along in AI adoption because 70%+ of hospitals use ambient listening
REALITY: Most organizations haven’t done the hard work of changing job descriptions, processes, and leadership models to match AI capabilities

The Equity Imperative

Kyu Rhee, National Association of Community Health Centers

Community health centers serve one in seven Americans, including one in three people in rural America. Yet less than 5% of their exam rooms have ambient AI, compared to 70% in hospital settings.

“We’re high trust, but not yet high tech,” Rhee acknowledges. The opportunity? The Rural Health Transformation Fund’s $50 billion could be catalytic in closing this gap.

NACHC is pioneering a Digital Health Formulary for safety-net providers, modeled after pharmaceutical formularies that evaluate clinical evidence and negotiate prices. “Health centers are overwhelmed. Everyone’s putting ‘AI sprinkles’ on their products. We need to create transparency about what works, for whom, and at what cost.”

Rhee’s patient-governed boards (51% of community health center boards are patients, by law) are demanding that AI be “responsible, ethical, and transparent.” And they’re asking the critical question vendors often dodge: Was this trained on data from communities like ours?

The vision? Techquity, using technology to advance health equity in the highest-need communities. “I believe in the next 20 years, community health centers will serve almost 100% of rural America. We need AI tools built for that reality.”

MYTH: AI is just for well-resourced academic medical centers
REALITY: AI has the potential to democratize access and reduce disparities because of its low marginal cost per person

Action Plan: The “Coal vs. Gold” Vendor Checklist

Ready to separate real AI solutions from hype? Here’s what these experts demand:

1. Require AI model cards/nutrition labels showing training data, performance metrics, and limitations
2. Insist on contracts ≤ 1 year in this fast-moving space
3. Demand third-party validation, not just vendor-reported results
4. Ask: “How does this perform on diverse populations?” especially those you serve
5. Tie vendor relationships to performance metrics with commitment to remediate when models underperform
6. Build advisory councils with external AI expertise to inform procurement
7. Start making mistakes—but on real solutions, not fool’s gold

What’s Next?

Looking ahead three years, these leaders see AI physicians managing chronic disease in underserved areas, techquity powered by federal rural health investments, and the inevitable obsolescence of fee-for-service models in an AI-optimized future.

Want the full conversation? Watch this complete Drive to ViVE episode to hear more about AI governance structures, the culture clash between tech and healthcare, and why ambient listening is just the beginning.

Join us at ViVE 2026 (Los Angeles, February 22 to 25) to continue this critical conversation. Register for ViVE 2026 and explore the AI Zone@ViVE2026 to see these principles in action.