	The One Hundred Year Study on Artificial Intelligence seems optimistic about the future of AI by 2030. However, the bright future may not be very easy to achieve given the constraints of privacy concern and tech-savviness of users. 
	Privacy may be the biggest obstacle that lays ahead of AI development. Since development of AI depends on running statistics models that based on huge amount of data, if privacy concerns limit the amount or quality of data for statistics models, then it will be hard to achieve accurate analysis, for instance, the healthcare analytics the report demonstrates. The report says that at the population level, AIs ability to mine outcomes from millions of patient clinical records promises to enable finer-grained, more personalized diagnosis and treatment (Stone el al., 26). However, how easy is it to obtain health data? One way to collect health data are from the automated capture of personal environmental data from wearable devices (Stone el al., 27). Yet, wearable devices currently are not cheap enough for everyone to afford them, but in order to fit the most reliable models, we need data that can represent the whole population. Will the price of wearable devices create selection bias in the health care models? 
      The best source of health records probably are the clinical records in health institution. However, how difficult is it to obtain these records from thousands of institutions? If owning a large amount of clinical data puts an institution in great advantage in terms of utilizing AI technologies to improve medical services, then it is likely that that a large institution will keep the data to themselves. Then it may be possible that the analysis using AI technologies ignore the population who go to small clinics or illnesses that do not require resources of a large hospital. Just as the report points out that he FDA has been slow to approve innovative software, in part due to an unclear understanding of the cost/benefit tradeoffs of these systems (Stone el al., 27), it may be reluctant for government to fully support AI development because the policy makers may concern about public opinion on privacy. 
      The report also enumerates list of thing AI could offer to help the including assistance in hearing, vision and mobility (Stone et al., 30). The list of helpful things AIs can do to help the elder certainly improve the elders quality of life where they have more control over their life without the help of other people, and thus they could have higher self-esteem. However, when people get old, their cognitive and physical capacity decline, and they take longer time to understand new products, especially those who have disability or diseases. Sometimes when it is too difficult to understand, like many cases with the computers and Internet today, people just get so frustrated that they give up on the new technology. Therefore, AI products need to fit the capacity of the targeted customers. Instead of the traditional way of customers learning how to use products, the user-friendly products need to understand precisely what elder customers need and initiate action to help them. To achieve this, it requires AI products to have very high intelligence, and we are far from that. Thus, in my opinion, the report is far too optimistic on what we can achieve with AI to help the elderly by 2030. 
