Should Doctors Use AI? My Honest Opinion
As someone who has been teaching and practicing pediatrics for years, I have witnessed medicine evolve in ways I never imagined during my medical school days. Today, I want to share my thoughts on something that is rapidly changing our profession - artificial intelligence in clinical medicine. This is not about robots replacing doctors. Rather, it is about how we can use AI as a powerful tool to become better clinicians while keeping the human touch that defines our profession.
The Current State of AI in Medicine
Let me start with some real numbers. According to recent data from August 2024, the FDA has authorized approximately 950 medical devices that use artificial intelligence or machine learning. By July 2025, this number has grown to over 1,250 AI-enabled medical devices. This is not science fiction - these are tools that are already approved and being used in hospitals and clinics across the country.
What surprised me even more is that 68% of physicians see at least some advantage to using AI in their practice, up from 65% in 2023. The medical community is warming up to this technology, and for good reasons.
Where is AI Being Used Right Now?
1. Radiology and Imaging
This is the area where AI has made the biggest impact. About 76% of all FDA-approved AI medical devices are related to radiology. These tools help radiologists detect abnormalities in X-rays, CT scans, and MRIs faster and more accurately. As pediatricians, we often order chest X-rays for pneumonia or abdominal scans for appendicitis. AI tools can flag concerning findings that might need immediate attention.
2. Clinical Documentation and Note Taking
This is where I see the most practical benefit for busy doctors. AI-powered ambient documentation tools can listen to our patient conversations and automatically generate clinical notes. These systems can save doctors hours of paperwork every day.
Imagine finishing a busy OPD and not having to stay back for hours completing charts. That is the promise of ambient AI documentation. The system listens, records, and generates organized notes in real time. You just need to review for accuracy.
3. Diagnostic Support
Here is something fascinating - a study published in JAMA in October 2024 showed that when doctors used AI diagnostic tools, their accuracy improved from 74% to 76%. But what caught my attention was that the AI alone performed even better at 90%. Before you panic, this does not mean AI should replace us. It means we need to learn how to work with AI effectively.
4. Cardiovascular Medicine
About 10% of FDA-approved AI devices are for cardiology applications. These include ECG interpretation tools, detection of heart failure from echocardiograms, and prediction of cardiac events. As someone who sees children with congenital heart conditions, I appreciate how AI can help identify subtle abnormalities in pediatric ECGs that might be easy to miss.
Practical Ways to Start Using AI
Based on my experience and discussions with colleagues, here are realistic ways doctors can integrate AI into their practice:
Start Small with Documentation
Begin with AI scribes or documentation assistants. These tools are less intimidating and provide immediate time-saving benefits. Most platforms integrate with existing electronic health records, making adoption easier.
Use AI for Second Opinions
Think of AI as a consultant who is always available. When you encounter a complex case or unusual imaging finding, AI tools can provide additional perspectives. This is particularly helpful in emergency situations or when specialist consultation is not immediately available.
Leverage AI for Patient Education
Some AI tools can generate patient-friendly explanations of medical conditions and treatment plans. This saves time and ensures patients receive consistent, accurate information.
Get Trained
Only 28% of physicians feel prepared to use AI effectively, despite 57% already using AI tools. This gap is concerning. Harvard Medical School now offers courses on AI in clinical medicine, and the AMA provides free CME modules on AI applications.
Important Limitations and Concerns
I would be dishonest if I only talked about the benefits. As doctors, we need to be aware of the limitations and potential risks.
Automation Bias
There is proven risk that doctors might over-trust AI recommendations. This can lead to missing things the AI missed or accepting flawed AI suggestions without proper verification. We must maintain our clinical judgment and not blindly follow AI outputs.
Limited Clinical Evidence
A study published in April 2025 analyzing 903 FDA-approved AI devices found that only about 56% had clinical performance studies reported at the time of approval. About 24% explicitly stated no clinical study was conducted. This means many AI tools we use may not have robust clinical validation.
Data Privacy and Security
AI systems need access to patient data to function. We must ensure these tools comply with privacy regulations and that patient information is protected. Always check if the AI tool you are using is HIPAA compliant.
Bias in AI Algorithms
AI systems are trained on historical data, which may contain biases. For example, if an AI is trained mostly on adult data, it may not perform well in pediatric populations. Less than one-third of clinical evaluations provided sex-specific data, and only one-fourth addressed age-related subgroups, according to the 2025 study.
The Future of AI in Clinical Practice
Looking ahead, AI will become more integrated into our daily practice. Here are some developments on the horizon:
Real-Time Clinical Decision Support
Future AI systems will analyze patient data in real time during consultations and suggest diagnostic tests or treatment options based on current guidelines and patient-specific factors.
Personalized Treatment Plans
AI will help create treatment plans tailored to individual patients based on their genetic makeup, lifestyle factors, and response to previous treatments. This is particularly exciting in pediatrics, where we often struggle with medication dosing and treatment responses in different age groups.
Predictive Analytics
AI tools will predict which patients are at high risk for complications or readmissions, allowing us to intervene proactively. This could significantly improve outcomes in chronic disease management.
Better Integration with Electronic Health Records
Currently, many AI tools work as standalone systems. Future development will focus on seamless integration with hospital information systems, making it easier for doctors to access AI insights without switching between multiple platforms.
My Personal Approach to AI
As a pediatrician and educator, I have adopted a balanced approach to AI. I use it as a tool, not a crutch. For documentation, I have started using an AI scribe that saves me about an hour each day. This extra time allows me to see more patients or spend additional time with complex cases.
For diagnostic support, I occasionally use AI tools for second opinions on imaging studies or unusual presentations. However, I always verify the AI suggestions with my clinical judgment and consult human colleagues when needed.
I have also incorporated AI discussions into my teaching. Medical students and residents need to learn how to work with AI effectively, just as they learn to use stethoscopes and interpret lab results.
Final Thoughts
AI in clinical medicine is not about replacing doctors. It is about giving us tools to be more efficient, accurate, and available to our patients. The technology is advancing rapidly, with physician adoption nearly doubling in just one year. However, we must approach it thoughtfully, understanding both its potential and limitations.
The doctors who will thrive in the future are not those who resist AI, but those who learn to use it wisely while maintaining the human elements that make medicine both an art and a science - empathy, clinical judgment, and the doctor-patient relationship.
Start small, stay curious, get proper training, and always put patient safety first. That is how we can harness AI to become better doctors without losing what makes us human.
About the Author: This blog reflects personal opinions based on current evidence and experience in pediatric medicine. The statistics and data mentioned are sourced from peer-reviewed journals, FDA databases, and medical association reports published between 2024-2025.
References:
- American Medical Association - Augmented Intelligence in Medicine (October 2024)
- FDA AI-Enabled Medical Devices Database (August 2024 - July 2025)
- JAMA Network Open - Study on LLM Diagnostic Tools (October 2024)
- Harvard Medical School - AI in Clinical Medicine Program (2025)
- JAMA Network Open - FDA-Approved AI Device Generalizability Study (April 2025)
- Nature Digital Medicine - FDA AI Device Analysis (July 2025)
- CNBC - AI Tools in Medical Practice Report (October 2025)