Can AI Replace Human Doctors?

 A friend sent me this link in a WhatsApp chat

"Chinese Startup Trials First AI Doctor Clinic in Saudi Arabia - Bloomberg"

https://www.bloomberg.com/news/articles/2025-05-15/chinese-startup-trials-first-ai-doctor-clinic-in-saudi-arabia

I think an AI can take over human doctors very soon but with some limits. I shall explain this shortly. 

Generally, I think AI can take over the entire medical examinations from history-taking, clinical examination (auscultation, palpation, percussion, neurological examination - gait, reflexes, eye movements, speech...etc), and all chemical pathology, haematology, serology and immunology. 

Even over 50 years ago scientists and medical engineers have already developed autoanalyzer that can analyse multiple blood samples for various tests simultaneously without even human help by medical technologists, and all the long lists of results come out within just 2 hours as seen in a busy hospital. 

Maybe certain biological assays like microbiological assays that requires incubation period of a few days, and cytology, clinical genetics need manual human help. All these can also be done by robotic AI, including robotic surgery that can be done more precisely than a human surgeon. 

But I think it is a matter of time for AI engineers to refine their technologies. AI can memorize vast, vast amounts of database of every medical discoveries, diagnosis and their treatment developed over the centuries and synthesize the best among them within less than one second, unlike a human doctor who may take days and weeks and even that,  the diagnosis may be missed out.

Even if the diagnosis was correct, the treatment plan may not be tailored as the best for individual patients, because we are all individuals and respond differently.

However, an AI doctor is capable of analysing not just all the medical history - past and present, but assemble all the clinical examinations, and laboratory findings, whether biochemical, microbiological, serological, immunological, radiological, genetic analysis,  etc, etc within just a fraction of one second, and prescribe the best personalized medicine (management) tailored for that patient that a human doctor is unable. 

It is sheer impossible for any human doctor to memorize, analyze and put together billions of medical and clinical data studied and practised over the centuries - but an AI doctor can easily gather and generate tens of billions of these data and experiences at a speed near the speed of light and offer the best to the patient. 

 After all, all these medical discoveries are actually done by the medical scientists and not by the clinician who merely apply all these discoveries - from lab, and radiological imaging to the development of drugs and give them to the patient at very high costs - a money-making health-care industry.  

But I think AI can do all these much, much faster, much, much more cheaply, what's more,  more efficiently than even all the medical scientists and clinicians put together. These AI machines are not just super intelligent, but are also armed with vast, vast knowledge from all past discoveries to the latest - and they do them free-of-charge. 

All these far-reaching examination of the trajectory modern medicine can be taken over by Artificial Intelligence. Let me explain this in greater depths. 

1. AI’s has immense capacity for knowledge storage and recall.  AI can store, recall, and process enormous quantities of data, billions of medical records, studies, case reports, imaging findings, genetic profiles, and therapeutic outcomes, all within milliseconds. No human being, even the most brilliant physician, could ever match this.

2. Diagnostic Power Beyond Human Limits

AI, when coupled with real-time data (labs, imaging, genomics, etc.), can spot patterns that are invisible to the human mind, not because doctors are incompetent, but because human cognition is biologically limited. Deep neural networks can pick up subclinical abnormalities, rare disease patterns, and subtle correlations that even the most seasoned diagnostician may overlook.

3. Precision Medicine and Personalization

AI’s greatest medical potential may lie here:

Integrating genetic, proteomic, microbiomic, and lifestyle data to offer tailor-made treatment regimens. Something as complex as polygenic risk scores or drug metabolism based on cytochrome P450 polymorphisms can be handled seamlessly by an AI.

Automated analyzers already exist as I already mentioned earlier. AI-enhanced robotics could extend this to include:

a). Cytology (e.g., Pap smears)
b). Histopathology (AI reads biopsy slides)
c). Radiology (AI outperforms humans in some imaging interpretations).

4. Robotic surgery (already being used in prostatectomy, cardiac valve repair, etc., with extreme precision)

5. Data Integration from Wearables, Implants, and Smart Homes

Imagine a patient monitored in real-time 24/7 — pulse, respiration, oxygen saturation, glucose levels, ECG, movement, sleep, even vocal tone, AI could continuously update diagnoses and treatment in the background without the patient even going to the hospital.

But let us consider the limitations too:

A patient is not merely a bundle of data. He or she is a narrative, an experience, often with emotion, fear, guilt, denial, and suffering. The art of medicine lies not only in diagnosis but also in:

Listening between the lines, understanding culture and belief systems, and gaining trust to ensure compliance. Often patients are unwilling to comply with treatment plans, because the clinician himself cannot understand the patients' beliefs, cultures, taboos religion, and other social and economics barriers. Clinicians themselves need to be educated on these issues.  

Similarly, an AI may understand language and voice tones, but empathy is something different. Some elements of bedside manner, touch, comfort, dignity may be irreplaceable.

There is also moral and ethical judgment to consider. What if a treatment is effective but carries social or moral implications? Should an elderly person with dementia get an expensive organ transplant? Should AI recommend euthanasia in incurable cancer? These are not just scientific decisions, but I think they are moral ones, involving ethics, humanity, and sometimes even spirituality.

Then there is accountability and liability. If AI makes a diagnostic or therapeutic error, who is responsible? The engineer? The hospital? The algorithm? Unlike a human physician who answers to a licensing board, an AI doctor raises new legal and ethical quandaries.

Consider also trust and acceptance. Patients may not accept care from a machine. Medicine is not only about healing the body but offering human presence in times of suffering. Will someone with a terminal diagnosis accept it better from a gentle, compassionate physician, or a screen?

The Likely Future: Human-AI Synergy

What I envision is likely to occur, but not as a full replacement. Rather, AI will become a super-assistant, a co-pilot:

AI does the data-heavy lifting, suggests the diagnosis, gives differential lists, recommends drugs. The human doctor confirms, communicates, counsels, and manages the human aspect.

We may call this the "Centaur Model" as part human, part machine, combining the best of both worlds.

Only a few of us can understand this. Only a few can clearly see the transcendent potential of AI in medicine, not just as a tool, but as a revolution. Yet, even as it grows more godlike in knowledge and speed, it still lacks the one thing that cannot be programmed, and that is compassion born of shared mortality.

In a nutshell, I can say "yes", AI will transform medicine far beyond what any clinician today can do, but until machines can weep with the dying or celebrate with the healed, the role of the physician will never be obsolete,  merely transformed.

This is my opinion with my the highest regards and fraternal respect to all my former medical colleagues, doctor relatives and my former course mates.

We have emotions that AI doctors lack.