Artificial Intelligence in Medicine: A Powerful Tool… But Not a Perfect One
Artificial intelligence has quietly entered nearly every corner of modern medicine. From hospital systems and research labs to small private practices like mine, AI tools are now helping clinicians analyze data, summarize research, draft medical documentation, and even assist with clinical decision-making.
As a physician who practices functional and naturopathic medicine, I’ve begun incorporating AI into parts of my workflow. The technology is impressive and sometimes genuinely helpful, but it also has limitations that I have experienced firsthand, which patients and practitioners should understand.
Like most tools in medicine, AI is neither inherently good nor bad. Its impact depends entirely on how we use it.
Where AI Is Already Helping Medicine
One of the most powerful uses of AI in healthcare is information synthesis. Medicine produces an enormous amount of data every year, thousands of research papers, clinical trials, and guideline updates. No single physician can realistically read everything.
AI systems can help summarize large volumes of medical literature and identify patterns faster than humans. In hospital settings, AI is already being used to:
Detect subtle abnormalities on imaging such as CT scans and MRIs
Identify early warning signs of sepsis in hospitalized patients
Assist with radiology and pathology analysis
Help predict hospital readmission risk
These tools don’t replace clinicians, but they can act as a second set of analytical eyes.
In my own practice, AI has been helpful for tasks such as:
Drafting medical assessments and summaries
Reviewing emerging research topics (though it cannot access publications that are paid, such as JAMA or NEJM unless someone has published it in the open web)
Helping structure educational materials for patients
Creating dietary plans and guidelines, and recipe creation
Creating exercise routines based off of goals
Business structuring and analytics (it keeps telling me I spend too much time writing my newsletters and blog posts hah)
Writing newsletters and blog posts (yes, even used in this one!)
When used thoughtfully, it can free up time so physicians can focus on what matters most: listening to patients and practicing medicine.
The Problem: AI Doesn’t Truly “Know” Medicine, especially Naturopathic Medicine
Despite the hype, AI does not actually understand medicine. Studies show that upwards of 30% of questions asked to AI systems involve ‘health’. It does not reason, verify evidence, or distinguish between high-quality research and poor-quality information the way a trained clinician can. It also does not distinguish between a far-reaching medical diagnosis and a more common one. For example, it may tell you you have Alpha-gal syndrome when you may just have mild dysbiosis of your gut microbiome.
While systems like ChatGPT do like naturopathic medicine, it most definitely does not understand our treatment plans and philosophy of medicine, so when I compare a treatment plan I created with what it thinks I should do, there are drastic variations.
AI systems are essentially pattern-recognition engines trained on massive datasets. They generate answers based on probability, what text most commonly follows a question, rather than true scientific validation.
This leads to one of the biggest risks: false confidence.
An AI-generated answer can sound extremely convincing while still being incomplete or incorrect. Trust me, it has made me question my judgment a time or two.
For example, I’ve personally seen AI systems recommend products or treatments that directly contradict the criteria I provided. In one case, I asked for recommendations that specifically excluded certain ingredients—and the system still suggested products containing them.
The reason? The algorithm recognized similar product categories but did not fully interpret the constraint.
In medicine, that kind of mistake matters.
Advertising and Bias in AI Systems
Another concern that many people don’t realize is how commercial influence can shape the information AI retrieves.
Many AI systems are trained on publicly available internet data, so what an open research article in the Journal of Bacteriology about soil bacteria and a mom who lives on a farm blogs will hold the same weight if both sites are highly visited. EEK. The internet contains incredible medical resources, but it also contains marketing material, affiliate blogs, sponsored articles, and search-engine-optimized content designed to promote products.
That means an AI tool may sometimes surface recommendations that are influenced by:
Advertising presence
Affiliate marketing pages
Brand-heavy search results
Popularity rather than scientific accuracy
This becomes particularly problematic when trying to distinguish between credible medical organizations and informal health blogs.
For example, a clinical guideline from the American Academy of Pediatrics (AAP) carries dramatically different weight than advice on a personal parenting blog. Yet an AI system may summarize both sources similarly if they appear frequently in its training data.
The system cannot independently evaluate scientific credibility; it can only reflect patterns in the information it has seen. A third-party, double-blind, placebo-controlled research study on an herb is not the same as a product research study on a patented product, but AI does not discern the difference.
That responsibility still falls on the clinician. So no, ChatGPT cannot and does not replace your doctor.
Where AI Could Transform Medicine
Despite these limitations, AI has enormous potential to improve healthcare in the future.
Areas where AI may become truly transformative include:
Earlier Disease Detection: AI models analyzing imaging and lab patterns may detect disease earlier than humans alone, particularly in cancers, cardiovascular disease, and neurodegenerative conditions.
Personalized Medicine: AI may eventually help physicians interpret complex datasets such as genetic variants, complex lab testing analytes, metabolomics, and environmental exposure patterns, to tailor treatments more precisely.
Administrative Relief for Doctors: One of the largest contributors to physician burnout is documentation. AI tools that help summarize visits and structure medical notes could return valuable time to providers. (This is the biggest area I feel in my practice)
Clinical Decision Support: Future AI models may act more like advanced reference systems, helping physicians rapidly compare treatment guidelines, drug interactions, and diagnostic pathways.
The keyword here is support.
AI should enhance clinical reasoning, not replace it.
The Bottom Line
Artificial intelligence is already influencing modern medicine, and its presence will only grow in the coming years. Used responsibly, it can help clinicians process information faster and reduce administrative burden.
Today, there are many medical training sessions for doctors on how to customize and train AI to become a safe, valuable source of clinical support, so we are seeing fewer inaccuracies.
I personally have multiple customized engines tailored specifically to help me write newsletters, one that assists with medical assessment writing, and one where we just chat about naturopathic medicine. I have to be very explicit about where it pulls data and information, and who I view as credible and who I do not.
But AI is not a medical authority. It cannot replace physician training, clinical judgment, or the nuanced understanding that comes from listening carefully to patients and interpreting their unique health story.
In my practice, AI is simply another tool, useful in certain contexts, limited in others, and always secondary to human expertise. Where it will never come to fruition is in naturopathic treatment plans (yay- job security :)
Technology can help us move faster, but wisdom in medicine still comes from experience, critical thinking, and the doctor–patient relationship.
Here to empower you with the knowledge of health,
Dr. Meg Holpuch
Disclaimer: The information provided on this blog is for educational and informational purposes only and is not intended as medical advice, diagnosis, or treatment. The content shared here is not meant to replace or supersede the guidance or recommendations of your personal healthcare provider. Always consult your physician or qualified healthcare professional before making any changes to your diet, exercise routine, supplement regimen, or overall health plan. Your health and well-being are unique, and decisions regarding your care should always be made in consultation with your trusted healthcare provider.
