Recently, the topic of #will artificial intelligence replace doctors in the future topped the Weibo hot list, causing widespread discussion among netizens.
AI technology was introduced to the medical field in the mid to late 20th century and began to be used to assist physicians in documentation of consultations or as a statistical analysis tool to improve diagnostic efficiency.
With the development of modern computer technology, artificial neural networks, simulated expert systems and other technologies have been researched and applied one after another, which has substantially improved the level of AI, and medical AI has become more intelligent and powerful.
Currently, AI is involved in many aspects such as assisted diagnosis and treatment, medical imaging, disease risk prediction, psychiatric analysis, rehabilitation treatment, etc. This has played a great role in improving the efficiency of medical resource allocation, reducing the cost of medical services and improving the level of medical services.
On December 7, 2021, Ping An AskBob "AI Doctor" won the world's first International Human-Machine Competition on Integrated Diabetes Management with a score of 92.4 vs. 89.5 for the international team of attending endocrinologists.
Currently, AI has a stronger advantage in both information storage and information processing capabilities compared to humans. So, in the era of metaverse, will AI really evolve into an independent being more powerful than doctors and thus replace them?
01 Al vs Doctor? An assistant or an opponent
Marx first proposed the idea that "technology is an extension of the human organism" in Capital. He believed that the invention of various technologies was essentially a human being's attempt to extend his own organs.
Ernst Karp, a technical philosopher contemporary of Marx, proposed the "organ projection theory". He believed that any technology is an imitation of the human organ, that is, technology is essentially the external materialization of the human organ.
The same is true for AI. Medical AI is an integration of various advanced technologies, an imitation of the doctor's brain, replicating and enhancing some of the structures and local functions of the doctor's brain thinking, and a tool to execute the doctor's will.
Medical AI's more massive storage, more subtle identification, faster screening, etc., are an extension of the doctor's organ. With AI, the doctor becomes a more powerful doctor.
And the organs, however extended, are only a part of the human being, and do not become a complete and independent person, and thus replace the person.
On an external level, medical AI can only process domain-specific tasks to solve part of the physician's job.
For example, the Intelligent Assisted Tumor Detection and Analysis Assistant for abdominal CT images can only detect occupying tumors, but is blind to obvious vascular endothelial cell tumors, and relying only on medical AI can easily lead to missed diagnoses.
On an internal level, the intelligence of medical AI does not surpass that of physicians. Because AI's learning ability is based on algorithms rather than social interactions, AI can only use the knowledge "fed" to it by humans, but cannot innovate knowledge on its own.
Moreover, medical AI cannot be a warm, emotional and caring doctor. After all, technology is not all about the doctor, and it is equally essential to handle the doctor-patient relationship with professional ethics, empathy and humanistic care.
For example, doctors often tell "well-intentioned lies". It is difficult for a medical AI to decide whether it should work with the patient's family to conceal a terminal diagnosis and tell a "good lie" on a case-by-case basis, as human doctors do.
As an adversary, AI with only technology does not replace a real doctor; instead, as an assistant, AI makes the doctor as good as a tiger.
02 AI + doctors, embracing the medical meta-universe together
Many studies have shown that neither doctors alone nor AI alone are as effective as AI combined with doctors.
For example, the Third People's Hospital of Chengdu uses AI, imaging physicians and AI combined with imaging physicians to diagnose lung nodules and compare the diagnostic data, respectively.
The results showed that the sensitivity of AI, imaging physicians and AI combined with imaging physicians in diagnosing malignant lung nodules was 93.3%, 78.5% and 98.6%, respectively, and the specificity was 34.8%, 79.7% and 79.7%, respectively.
The comparison shows that AI, together with imaging physicians, can reduce the rate of missed diagnoses and improve the correct diagnosis by jointly detecting lung nodules and benign and malignant diagnoses.
For example, Erden Medical Group Downtown Hospital used traditional manual, AI and the combination of both to analyze head CTA images of patients with intracranial aneurysms and compare the data, respectively.
The results showed that the sensitivity, specificity and accuracy of manual combined AI diagnosis of intracranial aneurysm were 95.92%, 92.42% and 95.39%, respectively, which were higher than those of single diagnosis; the false positive and false negative rates were 7.58% and 4.08%, respectively, which were lower than those of single diagnosis.
This also shows that the sensitivity, specificity and accuracy of manual combined with AI for intracranial aneurysm diagnosis are higher than single diagnosis, and the traditional manual combined with AI can reduce the leakage rate and misdiagnosis rate of intracranial aneurysm.
Therefore, instead of comparing who is more powerful between doctors and AI and deducing who will be replaced, it is better to make AI go to doctors, each taking the strengths of the other, and combining them strongly.
AI goes to the doctor, but also frees the doctor from inefficiencies. Not only does it improve the efficiency of consultations, AI can also help doctors spend less time on day-to-day administration, which can often take up to 70% of their management time.
With this time, doctors may continue to deepen their medical field and become more professional doctors; or they may start to explore interdisciplinary fields, move forward at the intersection of medicine and data science, move towards AI, and reach a deeper level of association with AI.
The purpose of creating medical AI has always been the same as the goal of physicians - for the sake of humanity.
The adoption of medical AI means the adoption of innovative and labor-saving support systems. The more it does so, the better it can serve patients, and the more it frees up doctors' energy and time to explore a new future of medicine and a new path for human development in the age of the meta-universe.
 YANG Tongwei, SU Yonggang, FENG Zhanqi. On the basic characteristics of medical AI and its basic concepts in practice[J]. Medicine and Philosophy,2021,42(14):16-19.
 Xu Hanhui. How AI doctors tell "good lies"--Thinking based on Confucian family orientation[J]. Chinese Medical Ethics,2021,34(1):12-16.
 Liu Na, Zhao Zhengkai, Zou Jiayu et al. Artificial intelligence-based assessment of the efficacy of chest CT lung nodule detection and benign and malignant diagnosis[J].CT Theory and Application Research,2021,30(6):709-715.
 Ye Y, Wang Z, Cao CJ et al. Clinical study of AI-assisted analysis of CTA images for the diagnosis of intracranial aneurysms [J]. Journal of Local Dissection Surgery,2022,31(2):128-131.
 Jian Cao. Cost, substitution and algorithms of medical AI [J]. China Hospital Director,2021,17(8):77-79.