How much AI does the health workforce actually need?
Is the future of healthcare robo-doctors or will we achieve to keep it human? And, if it is the latter, how much AI capacity will the human need?
This article is based on a talk I have given at the HIMSS25 in Paris, on 12th of June, 2025, with the title “AI capacity building for health workforce”.
In this article:
why vision matters
why AI is a symptomatic treatment, not a cure for healthcare systems
why AI capacity-building will not be needed as a specific effort in mid-to long-term
why the core skills needed to excel in health workforce will not change
We are in the middle of a massive global change, largely driven by digital technologies, especially artificial intelligence. With this shift, just about every field is faced with some big questions around transformation of labor market and jobs, and healthcare is not being spared: How will healthcare jobs evolve? What new skills will healthcare workers really need? And, perhaps the most sensationalized question of all: are AI-powered robots going to take over from human doctors?
Let's start with that sensational idea of a fully automated medical future. You often hear headlines about AI replacing all kinds of human roles, from teachers to doctors, artists to politicians ( yes, some actually envision AI to run countries ).
Bill Gates has declared in March 2025 that “within 10 years, AI will replace many doctors and teachers” and “humans will not be needed for most things”. Now, if Bill Gates can dream of it, the Chinese have already done it. Recently, there were reports from China on the first completely AI-based healthcare facility that runs without humans. Apparently, it is a facility that runs with 14 “AI doctors” and 4 “AI nurses” with the capacity of managing 3000 patients per day (though strangely, it's hard to find much follow-up detail on how those are actually doing). This kind of narrative, while it might be exciting for tech-enthusiasts, fundamentally misses the point of what healthcare truly is.
Healthcare, at its core, is a human endeavor. A healthcare worker’s is simply not a job you can fully automate, because healthcare is far beyond just biology. And contrary to the blatant misconception of some in the tech sector, medical questions do not always have simple black-and-white answers. The idea that every health query has an absolute answer that AI can deliver is a dangerous misconception. It actually reminds one of those old, outdated views of doctors as all-knowing figures who just tell patients what's best for them. (However, why AI cannot replace human healthcare workers is a topic I’ll save for another blog post. For now, let’s just agree that our vision for healthcare is not robocop… ehm… robodocs.)
A human-centered approach should recognize AI for what it is, a powerful tool. It’s there to help and enhance what human healthcare providers do. This view embraces AI’s huge potential benefits, but it also acknowledges that AI is a probabilistic machine, can and will make mistakes, and is meant to be an assistant. It is not some all-knowing, all-powerful entity, an uber-species, that will or should replace us.
First things first: Start with “why"
Any big investment, especially in public health, has to start with a clear "why." Our vision for the future will and should determine everything about how we build capacity – what we teach, who we teach, and what we hope to achieve. If our dream future is full of robot doctors, then sure, we should go ahead and invest in robots, not human training. But if our guiding principle is the health and well-being of people, then every bit of technology we bring in has to serve that ultimate goal. Technology should serve people, not the other way around.
So, our fundamental policy vision must be the starting point: affordable, high-quality healthcare for all, powered by the best science and technology available.
Next: Identify obstacles and how technology can tackle them
Once that vision is clear, the next crucial step is to identify the big challenges stopping us from getting there.
Healthcare systems are globally under pressure and given the current global conjuncture, the pressure does not seem likely to dissolve anytime soon.
The World Health Organization (WHO) predicts a critical shortage of 11 million health workers by 2030. This hit will be hardest in low and middle-income countries, which are already struggling with growing healthcare needs, partly due to climate change affecting ecosystems (just look at what COVID-19 showed us). A chronic lack of investment in healthcare education and training, plus shrinking public budgets and cuts to global health programs, makes the staffing crisis worse. We’re seeing total healthcare spending stagnate and budgets shrink as a percentage of GDP in many countries. On top of it, it remains hard to get healthcare staff to rural and underserved areas, and many healthcare workers from low- and middle-income countries are moving to higher-income countries.
The need for healthcare workers is growing though, with the aging population in many countries and the growing impact of climate change worldwide. Rising global temperatures contribute directly to heat-related illnesses and deaths, particularly among the elderly (whose numbers are increasing with the demographic shift). Altered precipitation patterns and increased droughts undermine food security and nutrition, which results in malnutrition and famine. Changes in temperature and humidity also expand the geographic range and seasonality of vector-borne diseases like malaria and Lyme disease, as mosquitoes and ticks expand to new areas. Increased frequency and intensity of extreme weather events such as floods and wildfires directly lead to injuries, fatalities, and displacement. Wildfires and ground-level ozone formation in hotter temperatures decrease the air quality, which in turn worsen respiratory and cardiovascular diseases. Lastly, the mental health burden of climate change is significant, with rising levels of anxiety, depression, and post-traumatic stress disorder. The resulting burden is disproportionately carried by those in underserved areas, with less human and financial resources and insufficient infrastructure. Ergo, the areas that have the highest gaps between health workforce supply and need.
In the face of all these complex, deep-rooted problems, AI can offer symptomatic relief, rather than a “cure”. And as any good doctor will tell you, if there is a cure, it’d be negligent to treat the disease only symptomatically. While AI won't cure the underlying issues like climate change, underfunding or workforce shortages, it can alleviate the burden and support our overwhelmed health systems while we work on long-term solutions.
Figuring out the “how” and the kind of capacity we need
With our vision clear and obstacles understood, we can strategically figure out how AI can actually help. AI’s most promising uses in healthcare are all about making human work easier and processes smoother. The most obvious low-hanging fruit is cutting down the admin work (which happens to take a substantial part of their working hours), gathering information, summarizing patient reports and records that can sometimes be hundreds of pages long. It can be used in early detection and warning systems to analyze large amounts of data in real time to spot worrying situations or subtle changes in a patient's health, as well. The nature of how direct healthcare is delivered, on the other hand, is not likely to change. We will have more sophisticated tools, but we will still need people to use, understand and interpret those tools.
We’re in a transition period. Rapid tech advancements can sometimes make it hard to see the long game. While specific AI training is crucial now, its value as a standalone "capacity-building" effort will change over time. In the medium to long term, being good with AI will likely be as fundamental as being computer literate is today. It will just be an assumed skill, built right into broader education and professional development, rather than a separate course for healthcare professionals.
The World Economic Forum’s Future of Jobs Report (2025) points at a similar direction. The top three skills expected to gain most importance by 2030 are technology-related: AI and big data, networks and cybersecurity, and technological literacy. This is no wonder in a digital world, digital skills will become core skills that are expected from everyone. Following the technological skills in the list are soft skills: creative thinking; resilience, flexibility and agility; curiosity and lifelong learning; leadership and social influence; and analytical thinking.
For healthcare workers, the most important among these skills are technological and scientific literacy; analytical thinking; and resilience, flexibility, and agility. While the specific technologies that the workforce needs to adapt to are changing (as they have done in the past too), the core skills needed to adapt to them remain the same. Knowing how to understand and use AI ethically is essentially no different from needing to know how to use computers and the internet responsibly.
Capacity-building is simple.
Once we grasp these factors, building AI capacity for the health workforce actually becomes quite simple. For everyone in healthcare, we need technology and science literacy. The focus should be on building a basic understanding of how AI works, its limitations, its potential pitfalls, and how to use it safely and ethically. For innovators and researchers in healthcare, which form a substantially smaller group, we need deeper expertise, as they will need a much deeper technical understanding in order to create new solutions using the technology. Their training thus should equip them to combine their medical expertise with advanced AI capabilities to create groundbreaking, patient-focused solutions.
Around the world, we're already seeing solid efforts in this direction. From big capacity-building drives in places like the UAE (for example, a Global AI Healthcare Academy there trained 4,000 healthcare professionals within a year, which is significant when you consider there are about 16,000 physicians in the country) to dedicated AI innovation centers in Singapore (which is the country to watch if you want a good example of how to progress in AI implementation). These efforts are essential to help our current workforce adapt to this brand-new technology and ensure it’s used responsibly and ethically. Down the road, though, this kind of specific, standalone AI training won’t be needed as much, because the new generations will be AI-natives.
Building AI capacity in healthcare is a dynamic journey. It should start with a clear, human-centered vision for the future, a targeted approach to training different groups of people according to their needs, and the foresight to understand that while it’s critical right now, AI literacy will eventually just become an expected, built-in skill in the not-so-far future.