Artificial intelligence has long been heralded as a transformative force in medicine. Yet, until recently, its potential has remained largely unfulfilled.
Consider the story of MYCIN, a “rule-based” AI system developed in the 1970s at Stanford University to help diagnose infections and recommend antibiotics. Though MYCIN showed early promise, it relied on rigid, predetermined rules and lacked the flexibility to handle unexpected or complex cases that arise in real-world medicine. Ultimately, the technology of the time couldn’t match the nuanced judgment of skilled clinicians, and MYCIN never achieved widespread clinical use.
Fast forward to 2011, when IBM’s Watson gained global notoriety by besting renowned Jeopardy! champions Ken Jennings and Brad Rutter. Soon after, IBM applied Watson’s vast computing power to healthcare, envisioning it as a gamechanger in oncology. Tasked with synthesizing data from medical literature and patient records at Memorial Sloan Kettering, Watson aimed to recommend tailored cancer treatments. However, the AI struggled to provide reliable, relevant recommendations—not because of any computational shortcoming but due to inconsistent, often incomplete, data sources. These included imprecise electronic health record (EHR) entries and research articles that leaned too heavily toward favorable conclusions, failing to hold up in real-world clinical settings. IBM shut down the project in 2020.
Today, healthcare and tech leaders question whether the latest wave of AI tools—including much-heralded generative artificial intelligence (GenAI) models—will deliver on their promise in medicine or become footnotes in history like MYCIN and Watson.
Anthropic CEO Dario Amodei is among the AI optimists. Last month, in a sprawling 15,000-word essay, he predicted that AI would soon reshape humanity’s future. He claimed that by 2026, AI tools (presumably including Anthropic’s Claude) will become “smarter than a Nobel Prize winner.”
Specific to human health, Amodei touted AI’s ability to eliminate infectious diseases, prevent genetic disorders and double life expectancy to 150 years—all within the next decade.
As the author of “ChatGPT, MD,” a nonfiction book on generative AI’s future in medicine, I admire parts of Amodei’s vision. But my scientific and medical background makes me question some of his most ambitious predictions.
When people ask me how to separate AI hype from reality in medicine, I suggest starting with three critical questions:
Question 1: Will the AI solution speed up a process or task that humans could eventually complete on their own?
Sometimes, scientists have the knowledge and expertise to solve complex medical problems but are limited by time and cost. In these situations, AI tools can deliver remarkable breakthroughs.
Consider AlphaFold2, a system developed by Google DeepMind to predict how proteins fold into their three-dimensional structures. For decades, researchers struggled to map these large, intricate molecules—the exact shape of each protein requiring years and millions of dollars to decipher. Yet, understanding these structures is invaluable, as they reveal how proteins function, interact and contribute to diseases.
With deep learning and massive datasets, AlphaFold2 accomplished in days what would have taken labs decades, predicting hundreds of proteins’ structures. Within four years, it mapped all known proteins—a feat that won DeepMind researchers a Nobel Prize in Chemistry and is now accelerating drug discovery and medical research.
Another example is a collaborative project between the University of Pittsburgh and Carnegie Mellon, where AI analyzed electronic health records (EHRs) to identify adverse drug interactions. Traditionally, this process took months of manual review to uncover just a few risks. With AI, researchers were able to examine thousands of medications in days, drastically improving speed and accuracy.
These achievements show that when science has a clear path but lacks the speed, tools and scale for execution, AI can bridge the gap. In fact, if today’s generative AI technology existed in the 1990s, ChatGPT estimates it could have sequenced the entire human genome in less than a year—a project that originally took 13 years and $2.7 billion.
Applying this criterion to Dario Amodei’s assertion that AI will soon eliminate most infectious diseases, I believe this goal is realistic. Today’s AI technology already analyzes vast amounts of data on drug efficacy and side effects, discovering new uses for existing medications. AI is also proving effective in guiding the development of new drugs and may help address the growing issue of antibiotic resistance. I agree with Amodei that AI will be able to accomplish in a few years what otherwise would have taken scientists decades, offering fresh hope in the fight against human pathogens.
Question 2: Does the complexity of human biology make the problem unsolvable, no matter how smart the technology?
Imagine searching for a needle in a giant haystack. When a single answer is hidden within mountains of data, AI can find it much faster than humans alone. But if that “needle” is metallic dust, scattered across multiple haystacks, the challenge becomes insurmountable, even for AI.
This analogy captures why certain medical problems remain beyond AI’s reach. In his essay, Amodei predicts that generative AI will eliminate most genetic disorders, cure cancer and prevent Alzheimer’s within a decade.
While AI will undoubtedly deepen our understanding of the human genome, many of the diseases Amodei highlights as curable are “multifactorial,” meaning they result from the combined impact of dozens genes, plus environmental and lifestyle factors. To better understand why this complexity limits AI’s reach, let’s first examine simpler, single-gene disorders, where the potential for AI-driven treatment is more promising.
For certain genetic disorders, like BRCA-linked cancers or sickle cell disease that result from a single-gene abnormality, AI can play a valuable role by helping researchers identify and potentially use CRISPR, an advanced gene-editing tool, to directly edit these mutations to reduce disease risk.
Yet even with single-gene conditions, treatment is complex. CRISPR-based therapies for sickle cell, for example, require harvesting stem cells, editing them in a lab, and reinfusing them after risky conditioning treatments that pose significant health threats to patients.
Knowing this, it’s evident that the complications would only multiply when editing multifactorial congenital diseases like cleft lip and palate—or complex diseases that manifest later in life, including cardiovascular disease and cancer.
Put simply, editing dozens of genes simultaneously would introduce severe threats to health, most likely exceeding the benefits. Whereas generative AI’s capabilities are accelerating at an exponential rate, gene-editing technologies like CRISPR face strict limitations in human biology. Our bodies have intricate, interdependent functions. This means correcting multiple genetic issues in tandem would disrupt essential biological functions in unpredictable, probably fatal ways.
No matter how advanced an AI tool may become in identifying genetic patterns, inherent biological constraints mean that multifactorial diseases will remain unsolvable. In this respect, Amodei’s prediction about curing genetic diseases will prove only partially correct.
Question 3: Will the AI’s success depend on changing human behavior?
One of the greatest challenges for AI applications in medicine isn’t technological but psychological: it’s about navigating human behavior and our tendency toward illogical or biased decisions. While we might assume that people will do everything they can to prolong their lives, human emotions and habits tell a different story.
Consider the management of chronic diseases like hypertension and diabetes. In this battle, technology can be a strong ally. Advanced home monitoring and wearable devices currently track blood pressure, glucose and oxygen levels with impressive accuracy. Soon, AI systems will analyze these readings, recommend diet and exercise adjustments and alert patients and clinicians when medication changes are needed.
But even the most sophisticated AI tools can’t force patients to reliably follow medical advice—or ensure that doctors will respond to every alert.
Humans are flawed, forgetful and fallible. Patients skip doses, ignore dietary recommendations and abandon exercise goals. On the clinician side, busy schedules, burnout and competing priorities often lead to missed opportunities for timely interventions. These behavioral factors add layers of unpredictability and unresponsiveness that even the most accurate AI systems cannot overcome.
And in addition to behavioral challenges, there are biological issues that limit the human lifespan. As we grow older, the protective caps on our chromosomes wear down, causing cells to stop functioning. Our cells’ energy sources, called mitochondria, gradually fail, weakening our bodies until vital organs cease to function. Short of replacing every cell and tissue in our bodies, our organs will eventually give out. And even if generative AI could tell us exactly what we needed to do to prevent these failings, it is unlikely people would consistently follow the recommendations.
For these reasons, Amodei’s boldest prediction—that longevity will double to 150 years within a decade—won’t happen. AI offers remarkable tools and intelligence. It will expand our knowledge far beyond anything we can imagine today. But ultimately, it cannot override the natural and complex limitations of human life: aging parts and illogical behaviors.
In the end, you should embrace AI promises when they build on scientific research. But when they violate biological or psychological principles, don’t believe the hype.
