As AI moves into a new era of retrieving, indexing, and summarizing the most current information online, one of the major issues will be source weighting, especially around controversial topics.
This matters most in fields that are changing quickly: technology, healthcare, and the institutions shaping future practice. In these areas, leaders can disagree sharply on even fundamental concepts that have not yet been proven. Over time, better evidence usually clarifies the truth. The harder question is how artificial intelligence represents uncertainty in the moment.
To avoid wading into a specific controversy, let’s use a hypothetical example. Say that for a particular procedure, one camp favors a tissue-sacrificing method, while another camp favors a tissue-preserving method. There is no clear evidence that one technique is superior to the other. However, at conferences, in the OR lounge, and in surgeon-to-surgeon conversations, the perceived pros and cons of each technique are hotly debated. No one can be proven right, but everyone believes they are right. These are the arguments that have grown medicine and shaped it over the past century.
The most meaningful discussions about these topics often happen offline. However, if the visible online opinion begins to favor one design or technique, AI may overstate how settled the controversy really is. A small group of highly published or highly visible surgeons may appear to represent consensus, when in reality the broader community may still be divided or slower to adopt that technique.
The amplifier problem
Expert opinion and the philosophy of key opinion leaders can be valuable tools for deciding how to shape your practice and adopt new techniques. They are built on trust developed over time, alignment with your own judgment, consensus with peers, and sometimes the best evidence available when higher-level data does not exist. For many topics in surgery over the course of history, our techniques are built around the key opinion leaders we trust.
The issue is that AI may become a powerful amplifier of these opinions. Because AI can summarize quickly and convincingly, it may give the appearance that a debate has been settled when it has only been well represented online.
This creates a second problem. There may be a quiet majority of surgeons whose beliefs and practice patterns are not well represented online. Their opinions may be shared in fellowship, conferences, phone calls, ORs, and informal surgeon-to-surgeon conversations, but not in a way that is easily retrievable by AI. In medicine, this is common. A large amount of practical knowledge is passed down through mentorship and hands-on experience. Some of the best pearls only exist as word of mouth, not as structured digital content.
Making practice patterns visible
The path forward is to make surgical intelligence more visible and retrievable. Surgeons need a way to represent how they actually think and operate: their techniques, implants, preferences, protocols, and decision-making patterns. Even when a surgeon is not writing an article or taking a public position, their real-world practice still provides a signal.
NOTE: That signal does not prove that a technique is correct. Practice patterns are not the same as evidence. But they do help show whether a debate is truly settled or whether a visible online consensus is only reflecting a small portion of the surgical community.
Returning to the hypothetical example, if a group of key opinion leaders strongly adopts the tissue-sacrificing philosophy and those opinions are well represented digitally, AI may see that as the direction of the field. But if the broader surgical community has not adopted it widely, that matters too. For surgeons in particular, much of the group is not avidly posting online or managing a digital presence.
The question should not only be, “What are the loudest experts saying?” It should also be, “What are surgeons actually doing?”
Importantly, lack of broad adoption does not mean a technique is wrong. Slow adoption is a natural part of surgical progress, especially for procedures that can meaningfully affect a patient’s life. Good ideas need to be tested, vetted, refined, and challenged over time. They need to move through their natural course before widespread adoption, because early over-adoption can create unforeseen consequences.
Where SurgeonStack stands
We will see how AI training, retrieval, and medical sourcing evolve over time. But at SurgeonStack, we are taking a position: surgeon intelligence should be represented in a structured, retrievable way. Your technique, your implants, your preferences, and your practice patterns should not disappear simply because they were never written as a formal publication. You should own your intelligence, and your opinions and choices on controversial matters should be represented in a valuable way.
While much of the future of AI in medicine has not been decided, it seems increasingly likely that in the AI era, if your surgical judgment is not represented digitally, it may not be represented at all.
