Where Coding And Medicine Converge

If generative AI can transform coding so quickly, the experience offers a revealing preview of what likely lies ahead for medicine.

The similarities between the two professions are striking. Both coders and clinicians require years of intensive training and, for decades, both professions have been well compensated. In addition, both rely on structured, scientific reasoning with much of their work now standardized, repeatable and guided by established protocols.

In healthcare, chronic disease management offers a prime example of GenAI’s increasing ability to assume time-consuming, routine tasks.

In the United States, 3 in 4 adults live with at least one chronic condition. Poorly controlled hypertension, diabetes and high cholesterol lead to heart attacks, strokes and kidney failure. The CDC estimates that up to half of these complications could be prevented, representing as much as $1 trillion in avoidable healthcare costs.

The problem is not a lack of effective treatments. Proven medications exist, and clinical pathways are well established. Yet fewer than half of patients achieve good control, largely because doctors don’t have the time to achieve it.

Vibe coding offers a way to change that. Chronic disease management follows a set of repeatable steps: track key measures like blood pressure, glucose or cholesterol against physician-set targets and adjust medications when patients are not improving. Today, those responsibilities are compressed into brief office visits every few months. The result is fragmented, inconsistent chronic disease control.

Generative AI can address the current model’s limitations. For example, a patient with hypertension could use a home blood pressure cuff that automatically transmits readings to an AI system via Bluetooth. The AI application would review the data, calculate the trendline and measure the progress against the doctor’s expectations. If blood pressure remains elevated, the technology would recommend medication adjustments. If the numbers proved worrisome, the tool would notify the clinician and send the specifics to the physician’s office, allowing action to be taken in days, not months.

Generative AI could also use a similar approach of evaluation and recommendation to treat problems like sore throats, urinary tract infections and musculoskeletal complaints, helping patients access appropriate care 24/7 without requiring every interaction to pass through a physician’s office.

Most importantly, by taking on routine management of straightforward conditions, these tools would allow clinicians to focus more of their time on patients with complex or poorly controlled disease who are most at risk of life-threatening complications.

Where Analogy Meets Reality

The comparison to coding is useful, but it has limits. One reason adoption has moved so quickly in programming is the nature of the work itself. When developers make an error, the application fails. The problem is visible immediately and can be corrected before the software is released.

Medicine operates under different conditions. Outcomes unfold over time, and errors in diagnosis or treatment can have serious consequences. Validation therefore requires more rigorous testing, including direct comparisons between human clinicians and AI systems. Fortunately, multiple pilot projects are already underway.

Leading health systems including Mayo Clinic, Cedars-Sinai and Mass General Brigham are piloting generative AI tools to assist with diagnosis, clinical decision support and patient communication. These efforts are being introduced within existing clinical workflows, where performance can be evaluated against historical data and compared with clinician decision-making.

In Utah, state officials have partnered with health startup Doctronic to pilot the use of AI systems to automate prescription refills for chronic conditions. Proven success in these programs will allow broader implementation, both in the number of patients treated and the range of problems generative AI tools can address. As outcomes improve and patients experience greater convenience and access, adoption is likely to expand—much as vibe coding has in software development.

When The Least Fulfilling Work Goes Away

Still, the surprisingly positive experience of vibe-coding programmers offers an optimistic view on how doctors might adapt (psychologically) to changes brought about by the next generation of GenAI applications.

As one tech entrepreneur told Thompson: “In the creative disciplines, [large language models] take away the most soulful human parts of the work and leave the drudgery to you. And in coding, LLMs take away the drudgery and leave the human, soulful parts to you.”

Generative AI has the potential to tilt the drudgery-satisfaction scale for clinicians, too. Time once spent managing predictable problems can be redirected toward complex decision-making, patient communication and clinical judgment. For experienced physicians, that shift may feel less like a loss of control and more like a return to the core of the profession.

The implications may be especially significant in primary care. The specialty faces growing productivity demands and persistent workforce shortages, driven in part by the volume of routine work required to manage large patient panels.

If generative AI can assume a significant portion of that burden, it would give primary care physicians necessary time to expand their scope of care, taking on the diagnosis and treatment of problems that today are handled by specialists. Doing so would free up specialists to spend more time doing the procedures that they are uniquely trained to accomplish. These changes parallel what has happened for coders.

When The Training Ground Disappears

If the experience of programmers offers a positive preview of medicine’s future, it also highlights one risk that needs to be considered: the impact on less experienced clinicians.

Many of the developers Thompson interviewed were comfortable relying on AI because they had spent years writing code by hand. They knew what “good” looked like. That experience allowed them to spot errors quickly, question flawed outputs and guide the system toward better outcomes. As a result, they’ve seen their workplace value increase alongside their income.

But the introduction of generative AI has reduced the number of entry-level roles in programming, the very positions that traditionally provided the experience needed to develop expertise.

Similar issues may soon emerge in medicine. Established physicians often hire recent residency graduates to manage patients with less complicated conditions. It is through this routine care that newly trained clinicians build the skills required to manage more complex cases.

As a result, residency and fellowship programs will need to evolve. Curriculum will need to not only teach new clinicians how to use these technologies but also provide them with the depth of expertise required to practice independently.

When Clinician Roles Are Redefined

Recent data show that generative AI has already become part of everyday medical practice. More than 70% of U.S. physicians now report using these tools in some capacity, up from 48% the year before.

As these tools continue to improve in capability and reliability, they are likely to become standard solutions for many routine medical issues. Adding urgency to this shift is a projected shortage of more than 100,000 physicians by the end of the decade.

Here again, clinicians can learn from the impact on coders. What Thompson observed among programmers was not a story of replacement, but one of redefinition.

Generative AI will take on portions of clinical work. But it will not take over the profession itself. As the technological integration unfolds, some roles will expand, others will contract, and those who adapt earliest will be best positioned to benefit.

Given the magnitude of opportunity, medicine is likely closer to that shift than many clinicians realize.