Consensus on delivery, divergence on content: physicians' artificial intelligence training needs in continuing medical education
A survey of 362 German physicians reveals that while there is a consensus on preferring short, online, and application-oriented delivery formats for AI training, significant divergence exists in content needs between current users and non-users, prompting a recommendation for a tiered continuing medical education curriculum delivered by physician chambers.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the medical world as a massive, high-speed train station. For decades, the trains (medical treatments) and the tracks (diagnostic tools) have been upgraded with incredible new technology. But recently, a brand new type of engine—Artificial Intelligence (AI)—has been bolted onto the locomotives. It's powerful, it's fast, and it's changing how the trains run. However, there's a problem: the conductors (the doctors) are being handed the controls before they've had a chance to read the manual.
This is the corner of science where technology meets human training. The paper focuses on a concept called "Continuing Medical Education" (CME), which is basically the mandatory "refresher course" doctors take to keep their licenses and learn new skills. Think of it like a driver's license update, but for the brain. The big question here isn't just if doctors are using these new AI tools, but how they are learning to use them safely. If a doctor uses a tool they don't understand, it's like trying to fly a plane by guessing which buttons to push; it might work for a moment, but it could be dangerous. The European Union has even passed a rule saying that anyone who uses AI at work must be trained on it, making this a legal and safety issue, not just a tech one.
So, what happens when you ask the doctors themselves what they need? A team of researchers decided to skip the guesswork and ask 362 practicing doctors in Germany directly. They wanted to know: Are you using AI? Did anyone teach you how? And if you want to learn, what does that lesson look like?
The results are a bit like a surprise party where everyone agrees on the cake but fights over the flavor. First, the "Are you using it?" part: more than half of the doctors (53.0%) are already using AI in their daily work. They are using it for everything from writing notes and organizing paperwork to helping with diagnoses and treatment plans. But here is the shocker: only 17.1% of them had ever received any specific training on how to use it. It's as if 53 out of 100 people are driving a race car, but only 17 have ever sat in a driving school. Most of the rest are just figuring it out as they go, often because their boss gave them the car, not because they asked for it.
The study also looked at why some doctors use AI and others don't. You might think it's about the doctor's age or how "tech-savvy" they are. But the researchers found that didn't matter much. Instead, it was all about the workplace. If a doctor's employer (the hospital or clinic) provided the AI tools, the doctor was over 11 times more likely to use them. It turns out that if you put the tool in their hands, they use it, regardless of whether they are young or old, or whether they feel confident with computers.
Now, let's talk about the training. The doctors were very clear on how they wanted to learn. They didn't want long, boring lectures that took up their whole weekend. They wanted short, snappy, online lessons—think 1 to 2 hours max—that they could do at their own pace. They wanted to learn how to actually use the tools in their real jobs, not just hear about the theory. They also overwhelmingly said they wanted their professional medical associations (the "chambers" that run the medical system) to be the ones teaching them, rather than the companies selling the software. They didn't trust the salespeople to teach them; they trusted their own professional leaders.
However, once they got to what to teach, the group split into three different camps based on their attitude toward AI:
- The Users: These are the people already using the tools. They wanted advanced, specialized training to get really good at it.
- The Interested Non-Users: These people aren't using AI yet but want to. They wanted to learn the basics of how the tools work and how to integrate them into their workflow.
- The Skeptics (Non-users with no intention): These are the people who have no plans to use AI right now. Surprisingly, they weren't just ignoring the topic. They were the ones who cared the most about the "rules of the road"—the ethics, the laws, and how to spot if an AI study was fake or biased. They also hated the idea of commercial companies teaching them, preferring independent, non-profit sources.
The researchers concluded that we can't just throw one big training course at everyone. Instead, we need a "tiered" system. Everyone starts with a short, shared foundation (the "intro" module). Then, depending on where they are on their journey, they move up to different levels: a "Trust and Rules" level for the skeptics, an "Entry into Application" level for the interested, and an "Advanced Use" level for the heavy users.
The most important takeaway is that the current system is broken. Employers are handing out AI tools without giving the training, and the doctors are mostly learning on the fly. The study suggests that the medical profession needs to step in, create these short, online, practical courses, and deliver them independently of the hospitals. This way, every doctor, whether they work in a high-tech city hospital or a small rural clinic, can get the training they need to stay safe and effective. It's not about forcing everyone to love AI; it's about making sure that if they do use it, they know exactly what they are doing.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.