Rejoinders to artificial intelligence and life in 2030

Recently, Stanford University published an article named Rejoinders to artificial intelligence and life in 2030 which summarizes the development of AI to date and makes many predictions based on the trend. As to the future, I have some different opinions:
First of all, the author thinks that city-dwellers will own fewer cars, live further from work, spend time differently and make traffic jams and parking challenges become obsolete, leading to an entirely new urban organization due to self-driving vehicles. The article assumes that people may rent autonomous driving cars and share with each other so that amounts of cars will decline. Since there are only 13 years left before 2030 and only semi-autonomous driving like Tesla’s Autopilot begins to mature, it seems difficult to see the sceneries described by author in 2030. And I think city-dwellers may own as same cars as nowadays because cars are rigid demands for work and life in North America. Compared to the public transportation, private vehicles are faster and more convenient and offer private room during driving. If people can bear the time waiting for other carpoolers and do not mind sharing one small room with others, they will choose carpool now. The decision on whether carpool or not has nothing to do with autonomous driving cars. Since amounts of cars may not decrease, traffic jams and parking challenges will still exist and trouble us in the future.

Secondly, the article is a little out of date on discussion of Healthcare analytics. Last year a Chinese scientific group consists of experts in Shanghai Jiao Tong University and Zhejiang University has developed a medical automated image interpretation system called ‘Doctor Alpha’ by using the machine learning model region convolution neural network. In a competition between Doctor Alpha and three best physicians, Doctor Alpha reads 300 magnetic resonance images in 23 seconds and its accuracy is 95.22% while human physicians spend an average time of 5 minutes to read 149 magnetic resonance images and human’s accuracy is only 93%. Besides Doctor Alpha, IBM has a similar system Watson. Since the technology of medical image processing has been developed, we do not need to invent another wheel. If an American company want to do some AI researches on medical imaging but is impeded by the Health Insurance Portability and Accountability Act, it can go to some countries or regions where do not strictly protect patients’ privacy as America does, like China.

Thirdly, the author says that robots may be able to deliver goods to the right room in a hospital, but then require a person to pick them up and place them in their final location. If you have watched videos published by Boston Dynamics, you will find robots’ hands are not as skillful as humans’ so that robots may break glass drug bottles and hurt the aged in the hospital.

Besides, the article asserts that Machine Translation technology will
also make it easier to translate educational material into different languages with a fair degree of accuracy. However, languages are different from one to another and there is no exact match between languages from different language families. So the machine translation may not be satisfactory and needs manual interventions. Books just translated by machine may be full of errors and could mislead students. Textbooks are so important for study that it is inappropriate to just use Machine Translation technology.

All in all, the author may be a little more optimistic about AI’s development and I believe there are lots of work remained for us to accomplish before the maturity of artificial intelligence.