ARTIFICIAL INTELLIGENCE AND LIFE IN 2030 discussed possible applications of artificial intelligence technologies in many domains, including the domain of healthcare system. The authors pointed out several bottlenecks of the development of applying AI in such domains: Outdated Legal regulations and incentive structures slowed down the development; Technical difficulties and risks slowed realization of AI's promise in healthcare; Data gathering process that might involves violation of privacy is also controversial. However, those obstacles can be removed, especially when political and economical impacts of applying AI in healthcare domain gains sufficient publicity in future. On the other hand, besides technical difficulties, people should also realize that the biggest challenge is to safely and widely provide healthcare to public through AI at lower cost. In other words, designing AI to detect diseases in cheaper ways, is more important than making it to run more accurate diagnosis.

All technologies in the domain of healthcare experienced obstruction of outdated regulation and incentive structures. When public safety is under consideration, it is natural for policy makers to be conservative on making changes. Just like and new drug would be forced to go through animal and clinical tests that last more than a decade, applications of AI in healthcare domain certainly will, and shall, experience rigorous and long term testing procedures. Its also rather better for the sake of public safety, to slow down reformations on regulations, than to change regulations with even smallest uncertainty and risks. Violation of privacy wont be a problem too. First of all, policy makers can make certain hash process mandatory to totally isolate the patient's medical data from their identity information. Secondly, privacy have already been violated in the name of public safety for many years (PRISM project for example) computer scientists should really just leave these questions to politicians and corporations. 

The motivation and resources to finally resolve the above problems wont and shouldnt come from academia. When AI can technically applied in healthcare domain, billions dollar worth of capital would flood into such area. Lobbyists would clear the obstacles. 

However, computer scientists should not focus on competing accuracy with medical researchers. Take using machine learning to detect alzheimer's disease for example. Today artificial neural network indeed did better job on detecting the disease based on MRI images. However, autopsy is still the only conclusive way. It is meaningless to spend too much effort on improving 0.1% of accuracy since currently there are more straightforward biological ways to make 100% correct conclusion. Since biological methods exists, it is not harmful to leave type I error and simply provide indication of high risk, rather than diagnosis. Computer scientists instead should focus on improving speed and lowering cost of AI so that more people can benefit from the technology and more capital can be drawn to the area to support future development.

Applying AI in healthcare domain would be crucial to national strategy, especially for countries with hundred millions people, and it will be the single most important part for any successful reformation on health care system. Medical use of AI would be the first time to truly industrialize the healthcare industry by making the productivity not constrained by human resources, because today, the capacity of health care system in a country is still limited by number of doctors and nurses. AI however can massively amplify the efficiency of a single health care providers, hence will be the ultimate solution. 

