General-purpose large language models outperform specialized clinical AI tools on medical benchmarks
   8 min read

AI is rewriting the rules of medicine in 2026, and the stakes could not be more personal — your genome, your cancer history, your body’s specific quirks are now data points that algorithms can actually act on. Fast Company’s deep look at AI-driven personalized medicine makes one thing unmistakably clear: the era of one-drug-fits-all is ending, and what’s replacing it is faster, stranger, and more powerful than most people realize.

The facts:

  • The University of Chicago Medicine appointed Josh Denny, MD, MS, as inaugural Vice Dean and Senior Vice President for AI, Precision Medicine, and Chief Data Officer, with his role beginning in November 2026.
  • Denny previously served as CEO of the NIH’s All of Us Research Program, described as the world’s largest integrated genomic and electronic health record resource.
  • Researchers at Sylvester Comprehensive Cancer Center, part of the University of Miami Miller School of Medicine, published a proof-of-concept blueprint for an AI companion designed to support cancer survivors between doctor visits.
  • UK-based Resurrect Bio uses AI to map interactions between plant immune receptors and pathogen proteins — an approach its CEO Cian Duggan compared directly to personalized medicine, according to AgFunderNews.
  • Hundreds of thousands of people finish cancer treatment each year and face a largely unsupported gap between clinic visits.

The Institutions Are Finally Catching Up

For years, personalized medicine was a buzzword that lived mostly in conference rooms and grant applications. Now, major academic health systems are building it into their org charts. UChicago Medicine’s appointment of Josh Denny isn’t just a hiring announcement — it’s a signal. When one of the country’s most prestigious medical institutions creates a brand-new senior leadership role fusing AI, genomics, and clinical data under one roof, that is institutional infrastructure being built in real time.

Enjoying this story?

Get sharp tech takes like this twice a week, free.

Subscribe Free →

Caucasian man smiling while using automated healthcare booth for blood pressure and health checks.

Denny spent over two decades connecting clinical records, genomics, and computation before landing at NIH to run All of Us, a program assembling the most comprehensive integrated genomic and health record database on earth. His move to UChicago in November 2026 tells you where the serious money and serious minds think the next decade of medicine lives. AI-driven precision medicine is no longer a research curiosity — it is a senior vice presidency.

Cancer Survivors Are the First Real Test Case

The Sylvester Comprehensive Cancer Center’s AI companion project targets one of healthcare’s most overlooked populations: the hundreds of thousands of people who finish cancer treatment each year and are then largely left to manage alone. Fatigue, anxiety, nutrition questions, insurance nightmares — all of it sits in the white space between appointments, where doctors aren’t present and Google is a genuinely terrible substitute.

Close-up of a hand holding a petri dish filled with various tablets and capsules on a neutral background.

The research team published a blueprint for how to build this kind of AI support responsibly. That word — responsibly — is doing serious work here. An AI that tells a cancer survivor the wrong thing about a drug interaction isn’t a minor UX failure. It’s a medical event. The fact that these researchers led with responsible deployment rather than rushing a product out says something about the maturity of this specific corner of health AI. Not every corner deserves that credit.

Personalized AI medicine works best when it fills genuine gaps rather than replacing human judgment. An AI companion available at 2 a.m. when anxiety spikes is not a replacement for an oncologist — it’s infrastructure that currently doesn’t exist at all for most patients. That distinction matters enormously, and most coverage of health AI buries it.

Personalized Medicine Is Spreading Well Beyond Human Patients

Here’s the part that most health tech coverage completely ignores: the logic of personalized medicine is already migrating into agriculture. According to AgFunderNews, UK startup Resurrect Bio uses AI to map interactions between plant immune receptors and the proteins pathogens deploy to suppress those defenses. CEO Cian Duggan describes the approach explicitly as “very targeted, almost like personalized medicine” — and the analogy holds. Instead of a broad chemical attack, Resurrect Bio’s FloraFold platform identifies the exact molecular weak point where a crop’s immune system has been outwitted, then guides gene edits to restore the plant’s own defenses.

This mirrors what’s happening in human oncology — stop blanketing the system with blunt instruments, identify the specific failure point, intervene precisely. The same AI-driven pattern recognition that could match a cancer drug to your specific tumor biology is being applied to make a wheat crop resist a fungus. Scientists are rethinking biology at the production level across industries, and personalized medicine AI is part of the same broader shift in how we deploy precision tools against biological complexity.

The Equity Problem Nobody Wants to Say Loudly

Here’s the contrarian read: personalized medicine AI, for all its genuine promise, currently risks becoming the most sophisticated health inequality machine ever built. Genomic databases — including the NIH’s All of Us program — have historically underrepresented Black, Latino, and Indigenous populations. An AI trained on incomplete data doesn’t become personalized medicine. It becomes highly optimized medicine for the people who were already best served by the system.

Josh Denny’s work on All of Us has actively tried to address this by building a more representative dataset, and that matters. But the problem isn’t just data collection — it’s who gets access to the outputs. Automation historically expands access in some sectors while concentrating benefit in others. Personalized medicine AI will follow the same pattern unless health systems, payers, and policymakers build equity in from the architecture level — not as an afterthought bolted on after the technology is already deployed and profitable. Right now, there is no widespread evidence that is happening fast enough.

If your health system isn’t already building AI infrastructure that accounts for your specific biology, your history, and your gaps in care, it is falling behind — and the distance between institutions that get this right and those that don’t is about to become a measurable gap in patient outcomes.

Source: Fast Company | See how media institutions are restructuring around new infrastructure

Watch the Breakdown

Sources

Need web hosting for your next project?

Fast, affordable hosting — get your site online in minutes.

Get Hostinger

Affiliate link — we may earn a commission at no extra cost to you.

Charles is the founder of Everyday Teching and Town Talk App LLC. A tech enthusiast, entrepreneur, and contrarian thinker who believes most tech coverage is broken. Everyday Teching exists to fix that...

0 0 votes
Article Rating
Subscribe
Notify of
guest

0 Comments
Newest
Oldest Most Voted