Aneesh Chopra: AI as the missing piece for healthcare productivity, drug discovery, and value-based care
Aneesh Chopra, the first chief technology officer of the U.S. and current chair of the Arcadia Institute, is clear on one point: this is not a bubble. He's bullish on AI's practical potential to fix healthcare's productivity slump, accelerate research, and make value-based care real.
Why he's bullish: the "last mile" of digitization
Healthcare poured billions into digitization, yet productivity declined. Staff spend more time per task, not less. Chopra sees AI as the missing layer that converts data exhaust into usable workflows, decisions, and outcomes. Think of it as finally cashing the check EHRs wrote a decade ago.
Yes, there's hype - and it's a distraction
AI won't fix broken incentives. Deployed inside fee-for-service, it's already fueling an arms race: upcoding on the provider side and automated downcoding on the plan side. That's inflationary and creates deadweight loss, not better care. The real gains come when AI lands in models that reward outcomes, not paperwork.
Value-based care is the unlock
Chopra points to Medicare's market-making role. Models that pay for outcomes - rather than stacking CPT codes - create room for AI to support clinical use cases that actually improve lives. He highlights the ACCESS approach as a pure form of performance-based payment: one code, half for treating the patient and half for delivering the outcome. You're either in that world or in fee-for-service; straddling both won't work.
Critiques that ACCESS pays "too little" miss the point, he argues. The intent is to stop layering payments on top of a flawed system and instead reward outcome delivery at scale. That requires policy, workflow redesign, and clinicians and plans aligned to the same scoreboard - not just "add tech."
CMS Innovation Center continues to evolve models that push in this direction. For leaders preparing to make the shift, the early mover advantage will go to organizations that rebuild processes around outcomes and real-time data.
Project Stargate and the promise for drug discovery
On the research front, Chopra is optimistic about large-scale private initiatives like Project Stargate, which the administration has celebrated but is funded by industry. The vision: massive AI infrastructure that can test and iterate hypotheses far faster than today's grant-driven cycles.
He sketches a vivid scenario: a researcher with a handful of ideas, now amplified by hundreds or thousands of AI agents running experiments, ranking pathways, and surfacing promising targets. That could mean faster routes from biomarker signals to therapeutic candidates and earlier reads on what might slow disease progression.
For context on traditional grants, see NIH's R01 overview: NIH Research Project (R01) grants.
What health systems and plans should do now
- Pick outcome-first use cases: Prior auth relief and documentation clean-up are fine, but prioritize clinical impact areas (e.g., readmission reduction, diabetes control, behavioral health engagement).
- Design for value-based payment: Build AI into care pathways tied to measurable outcomes, not CPT optimization. Make sure contract terms reward what your models improve.
- Tighten data interoperability: Stand up pipelines that pull EHR, claims, SDOH, RPM, and patient-reported data into a consistent layer. TEFCA participation and FHIR-first workflows will pay dividends.
- Close the loop at the bedside: Integrate AI into clinician workflow with clear recommendations, uncertainty flags, and a fast path to override. Tools must save time, not add clicks.
- Measurement and governance: Track impact on outcomes, costs, equity, and safety. Establish model oversight for drift, bias, and failure modes. Document intended use and monitor ROI monthly.
- Build patient trust: Be transparent about AI-assisted decisions, data use, and consent. Offer clinician review and a clear escalation path for complex cases.
- Prepare the workforce: Upskill clinicians, care managers, and rev cycle teams to work with AI copilots. Define new roles in model stewardship and clinical informatics.
Signals to watch
- Payer adoption of outcomes-only contracts: Follow which plans commit to CMS-aligned models and how they price risk.
- Regulatory clarity on AI in care delivery: Guidance on transparency, validation, and safety will shape deployment speed.
- Research throughput: Early wins from large AI infrastructure - more validated targets, faster trials, better stratification - will hint at timelines for new therapeutics.
Chopra's bottom line: AI works when incentives, interoperability, and clinical workflow line up. Get those right, and we'll see real productivity gains and better outcomes - not just smarter billing.
Related reading: AI for Healthcare
Your membership also unlocks:
