AI that Actually Saves Lives
*This article was originally published on Healthbeat*
Dr. Dave A. Chokshi, who will be a keynote speaker during The New York Academy of Sciences’ upcoming The New Wave of AI in Healthcare conference, shares his perspective about the potential of artificial intelligence to improve health.
Published April 13, 2026
By Dave A. Chokshi, MD
According to tech leaders, artificial intelligence will be transformative for our health. AI could give us “the next 50-100 years of biological progress in 5-10 years,” declared Dario Amodei, CEO of Anthropic. “Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer,” enthused Sam Altman, CEO of OpenAI.
Back in the clinic, our enthusiasm is more guarded. Part of it is the checkered history of health technology, too often extracting humanism in medicine for the pursuit of profits. Another hesitation is because so often, in the real world, the problem is not discovery but delivery. Cures that don’t reach the patients most in need; prevention that never makes it into practice. But what if AI could actually help us change that?
Take one of my longtime patients, whom I recently saw on his 45th birthday. We commiserated about the travails of middle age (I’m also turning 45 this year), and celebrated some progress he had made with his mental health in the last few months. And — with colon cancer affecting more young patients — I recommended a colonoscopy.
He acquiesced to a stool test, which screened positive, but declined the follow-up colonoscopy that might save his life. When I reached him to understand why, I learned that he had lost his job and his health coverage, and felt he had bigger fish to fry. “I’m feeling fine, doc” he said to me. His reassurance just reinforced my own powerlessness to reassure him.
My patient is far from alone: Roughly half of Americans with a positive stool test don’t complete follow-up within six months, despite the proven benefits of early detection and treatment.
This yawning gap repeats itself across American medicine. Less than half of patients with high blood pressure are on an effective regimen for it. Only about a third of patients who qualify for the medicines that could cure them of hepatitis C receive them. Fewer than 1 in 5 patients who would benefit from therapies like buprenorphine for opioid addiction get treated.
Even as we discover more effective treatments, we fail to deliver them to those most in need.
AI Can Help Patients Get Through Complex Process to Care
The discovery-delivery gap is not new. Three decades ago, health care pioneer Don Berwick highlighted “the gap between what we know and what we do.” But we now have a realistic opportunity to fix this failure, by thoughtfully harnessing AI to save lives.
Take my patient’s positive screen for colon cancer. Several steps are required after the stool test: notifying the patient, arranging scheduling, ensuring coverage, explaining the bowel prep. Sometimes the prep isn’t adequate, and we send patients back to square one.
The complexity of this process explains why many health systems employ patient navigators, and why people with greater means seek out concierge care to cut through the thicket. Artificial intelligence could help scale this kind of personalized, persistent engagement. Imagine a system that calls my patient in his native language, helps schedule his colonoscopy, and answers questions about the bowel prep. Nurses and social workers can focus their precious time on more complicated cases, like helping my patient enroll in Medicaid.
A test for AI apps is whether they enrich the human relationships at the heart of exceptional care. Too often, health technology has eroded relationships: reducing eye contact, depersonalizing interactions, making medicine transactional. Patients become less inclined to trust recommendations. In this environment, trusted intermediaries like community health workers become more important, not less, and technology must work for them.
Human relationships have always been a key to bridging innovations, like vaccines and cures, with implementation. The United States tamed tuberculosis, for example, through painstaking efforts to find and monitor cases in the community and provide subsequent support for completing treatment regimens.
AI Can Help Cures Reach Patients Who Need Them
Today we have a chance to eliminate modern scourges like hepatitis C, thanks to new therapies like curative antivirals. Hepatitis C is an insidious disease: Tens of thousands of Americans each year progress toward cirrhosis, their eyes yellowing and toxins clouding their minds as the liver fails. Yet a simple 8- to 12-week course of pills can eliminate the virus from the body. AI could help ensure that such breakthroughs actually reach patients.
Casefinding is especially important when people can transmit infection without knowing they are infected. Yet our tools for identifying these “silent carriers” are often blunt. Today, the national recommendation for hepatitis C is to screen every American between 18 and 79 at least once with a blood test. By contrast, a machine-learning algorithm tested in Israel was 100 times more efficient: yielding one diagnosis for every 10 tests instead of one for every thousand.
A similar opportunity exists for HIV prevention. As powerful new tools like the long-acting injection lenacapavir emerge, AI could help ensure that protection reaches those at highest risk.
Eliminating diseases like HIV and hepatitis C, just as we did for smallpox and polio, is within our reach — though it will take more than just technology.
What AI can’t do is solve our all too human problems, like the fact that high prices preclude broad access to some of our most consequential breakthroughs, from antivirals for hepatitis C, to the $28,000-a-year lenacapavir, to the new GLP-1 medications for diabetes and obesity. For this, we will have to work through our messy human systems to forge a better path.
Mere mortals have done it before. In Louisiana, a so-called “Netflix model” facilitated expanded access to hepatitis C medicines for a flat subscription fee. And the Congressional Budget Office estimated that a national subscription model for hepatitis C would save the federal government approximately $6.6 billion over 10 years, in part due to costs averted from hospitalizations and intensive care that would otherwise be borne by Medicare. We can save lives and money.
My patient ended up completing his colonoscopy after our social worker helped find health coverage he could afford. They found a small polyp that was precancerous, and he and I both breathed sighs of relief. But he’s due for another colonoscopy in five years. I hope we can make a century of medical progress by then. At the least, we can make it easier for patients like him to receive the care we already know saves lives.
Dr. Dave A. Chokshi will be a keynote speaker during this year’s The New Wave of AI in Healthcare conference which will take place May 12-13 at the Icahn School of Medicine at Mount Sinai. Learn more and reserve your spot at this impactful conference.
About the Author
Dr. Dave A. Chokshi is a physician at Bellevue Hospital and Sternberg Family Professor at the City College of New York. Previously, he served as health commissioner of New York City.
