The AI Paradigm: A Doctor’s Perspective

Healthcare is undergoing its most significant technological transformation in decades. At the center of this shift is generative AI (GenAI), offering scalable solutions to one of the sector’s greatest pain points: unstructured data. From physician enablement to drug discovery, the adoption of GenAI tools is accelerating daily. We are witnessing the beginning of a new digital era—one that is already reshaping how we practice medicine.

From the Exam Table to the Senate Floor: A Lifetime of Data-Driven Care

This is not the first time I’ve seen technology promise transformation. I’ve been working with medical data since the day I entered medical school over 50 years ago. My first scientific paper, written during those early years, involved collecting and analyzing vast amounts of data from bench-side experiments to study how papillary muscles contract under low oxygen conditions. That early experience planted a seed. Over the decades that followed—in academic medicine, in surgical practice, in policymaking—I’ve come to believe deeply in the power of data, if used wisely, to reveal truth.

AI is not a novelty. Its applications in healthcare date back to the 1970s when I as in pre-med and medical school with early clinical inference systems like MYCIN from Stanford. Over time, machine learning quietly found its way into medical imaging, insurance claims processing, and triage. But legacy IT systems and fragmented infrastructure kept it from reaching scale. That’s what makes GenAI such a powerful development. Since it burst into public awareness in 2022 through ChatGPT, it has offered a realistic path to address what I see as healthcare’s most systemic shortcoming: our inability to structure and act upon the vast amounts of data we already have and are accumulating everyday.

Healthcare generates a third of the world’s data—and it’s growing at 47% annually. Yet 80% of that data is unstructured. Across payers, providers, and pharmaceutical companies, years of uncoordinated and unconnected digitization have created rigidly siloed systems. Unlike other fields like fintech or logistics, where data flow freely, healthcare’s information remains locked in electronic medical records, PDFs, faxes, and call logs. We sit on a mountain of potential insight, but we’ve lacked the tools to access it—until now. As a transplant surgeon, I know firsthand the consequences of not having complete information at the point of care. I’ve made decisions in real time that would have been clearer, faster, and safer with better data access. The arrival of these new tools is why I am more excited than ever.

GenAI Arrives: Why This Moment Is Different

GenAI is proving to be a unique fix, and here’s why. It excels at structuring, analyzing, and interpreting unstructured data. It is already demonstrating success in automating patient communications, streamlining administrative workflows, aggregating fragmented patient records, identifying novel protein bonds for drug development, and more. And it’s arriving just in time. Our system is stretched thin.

Here are the four big challenges (and opportunities) I see:

1. We don’t know how to monitor and engage patients effectively. Fragmented data makes medicine reactive and intuition-based rather than proactive and evidence-based. My dad’s (a family doctor) day of a simple black doctor’s bag and all the knowledge in his head are ancient history now.

2. Clinicians and administrators spend far too much time manually sorting and reviewing information—time that should be spent with patients. Frustrations, burn out, and waste are the product.

3. Traditional business models are breaking down. One in five health systems has an operating margin below five percent.

4. Therapeutics work best when they are more personalized, requiring more complex and interconnected data to develop and commercialize.

Health services, technology firms, and life sciences companies are embracing GenAI at a speed I haven’t seen before in my lifetime in healthcare. My perspective today is the venture capital world, so I see daily the surging swell of interest and proposals based on AI coming though the door. While the industry is still in its very early days, every healthcare organization is exploring how to apply GenAI. Some are waiting for legacy systems to adapt; others are reinventing infrastructure from the ground up and building robust libraries of AI tools. The pace of improvement in foundation models from OpenAI, Anthropic, Grok, and others is astounding. Engineers are creating smarter tools by the hour. We’re all still learning—but no one doubts the revolutionary impact GenAI is poised to make.

The Opportunities (and the Obstacles) Right in Front of Us

In the near term, the greatest opportunity lies in integrating GenAI into existing systems to automate language and image-based tasks:

• Extracting data from medical records and claims

• Ambient note-taking

• Automated call center and triage

• Medical coding

• Prior authorization review

From what I am seeing as I interact with the innovation world, the clinical space, and the policy arena, I believe there are three sectors that stand to benefit the most:

1. Technology companies: Historically limited by the need to configure products for vastly different systems, they can now leverage AI to normalize and aggregate disparate data across formats.

2. Business services: Call centers and claims adjudication—in the past, fully manual—can be significantly automated with AI-driven tools, enabling teams to focus on more nuanced work.

3. Clinical services: Providers, especially nurses and doctors, and back-office administrators spend too much time searching for or submitting information. AI can coordinate care more efficiently, before and after patient visits, giving clinicians more time with patients. It’s what they want and the patient deserves.

With any new technology, as the progress unfolds the challenges become more apparent. The obvious ones now that we are grappling with now include:

• Change management—ensuring clinicians and staff adopt AI tools once implemented.

• GenAI’s imperfections—it can’t yet be fully trusted in deep clinical workflows.

• Safety and security concerns, especially in diagnostics, are slowing adoption.

• Fragmented IT infrastructure—differences in syntax and format still require significant fine-tuning for scalability.

A Glimpse Ahead: Reshaping the Future of Medicine

Looking further ahead, the horizon is both exciting and uncertain. It will take a deep understanding of policy, continual innovation in model development, and hands-on experience with healthcare delivery to fully realize AI’s potential. All parties need to be a the table and engaged. But I see several promising long-term shifts:

AI-Native Infrastructure — Just as the cloud matured into a universal platform, we’ll see the emergence of enterprise-wide solutions: AI-specific storage systems, organization-based foundation models, and optimization tools.

Therapeutic-Forward Care — Biopharma is investing heavily in AI to accelerate drug discovery. As clinical trials become faster and more efficient, the need for new services models to support personalized medicine will rise.

Diagnostic Support — GenAI’s use in diagnostics, already prevalent in fields like imaging, will expand as models become more accurate, reducing diagnostic errors (thus pain and expense) and supporting clinical decisions for better outcomes.

In my time as a surgeon, scientist, and policymaker, I’ve seen many waves of innovation. But few hold the promise of GenAI. It is not a magic wand, but it is a powerful tool—one that, if deployed wisely, will make medicine much more efficient, much more effective, and much more humane.

We have the opportunity—and the obligation—to shape this future thoughtfully. It will require the participation of all of us from many different perspectives. Let’s use it to empower doctors, support patients, and bring clarity to the data that until now has long eluded us. That’s how we’ll together turn the AI paradigm into a healthcare revolution.