Trust doctors or AI? this is a problem

According to a CCTV news report on November 3, not long ago, a parent took his 3-year-old child to the Department of Pediatrics at the First Affiliated Hospital of Zhejiang University of Traditional Chinese Medicine. The doctor considered that the child had asthma complicated by a lung infection and recommended that the child undergo a chest X-ray and other examinations. However, the parent refused to accept further examination on the grounds that the diagnosis given by the AI ​​was different from the doctor, and left with the child. But within a few days, the parent brought his child to see a doctor again. By this time, the child had developed severe pneumonia.

Nowadays, asking AI beforehand has become a choice and habit for many people. Some AI tools, including interactive question and answer, can indeed provide users with detailed analysis and satisfactory suggestions. Many doctors also affirm the ability of AI in seeking medical advice. However, "when AI recommendations are different from doctor diagnosis, who should be listened to" has become a controversial topic. In the above case, the patient's parents did not believe that the doctor believed in AI, which led to the worsening of the condition. This is a case worthy of attention.

Doctors have both professional qualifications and clinical experience, as well as "endorsement" from the hospital. It is said that patients should trust the doctor. However, many people are confused by the seemingly "omnipotent" and "incomparably correct" appearance of AI. When doctors give different diagnoses, it is difficult for some patients to change their blind faith in AI. In fact, AI technology is still immature today, and there may be problems such as hallucinations and algorithm bias.

Not trusting doctors but only trusting AI may not only "lay mines" for life and health safety, but also bring new challenges to the traditional offline medical treatment model and doctor-patient trust. In addition, this phenomenon also reflects that the current interactive question and answer AI tools are insufficient in warning of safety risks when answering questions involving life and health. For patients, medical institutions and related AI tool operators, the above cases serve as a warning lesson.

Patients need to realize that AI has limitations, and not all AI questions can give correct answers. If you blindly trust AI, at least it may delay your condition and increase medical costs, or at worst, you may face the risk of loss of life and health. If the public needs to improve their self-cultivation in the AI ​​era, then it should not only improve their ability to use AI, but also improve their ability to understand the limitations and risks of AI. Medical institutions should also reflect on why some patients do not believe that doctors believe in AI. This may be related to the highly deceptive "omnipotent" illusion created by AI. In addition, I am afraid it is also related to the phenomenon of over-examination and over-treatment that still exists in some places and medical institutions.

As for AI operators, they must not only promote their own products with a rigorous attitude and standardized publicity, but also continue to polish their products to solve problems such as AI illusion and algorithm bias, improve service quality, and reduce error rates. They must also remind users in various ways that AI recommendations are only for reference and have hidden risks, especially questions and answers involving life, health, and property safety.

From a regulatory perspective, it is necessary to further strengthen the value of doctors in a problem-oriented manner and clarify the auxiliary role of AI. Last year, the "Implementation Opinions on Promoting and Standardizing the Development of "Artificial Intelligence + Medical and Health" Applications jointly issued by multiple departments proposed "intelligent-assisted diagnosis and intelligent-assisted decision-making." In addition, doctors' dominance and decision-making power in diagnosis and treatment must be strengthened, and the common sense of "doctors as the mainstay, AI as the supplement" should be publicized and popularized to the public, and supervision of AI's non-standard services involving life and health should be strengthened.