Things will get more and more intelligent, and people will eventually get used to such intelligence. Currently, we are at a rudimentary stage (play music, tell me the time). The next stage is to add second-order inference of customers’’ intention.
Small-change and trust is the key. The product intelligence will be added incrementally and gradually. We will start small by adding basic functionalities like information assistance, and then we will gradually add higher-level assistance when people get used to the current level of intelligence. Trust is the cornerstone for all these to happen.
The ultimate goal of the cloud is to convert the computational resource into a type of utility. To meet that goal, three things must be met:
- The resource must be universally accessible; you should be able to access the cloud anywhere, with any device, at any time.
- The resource must be elastic and cheap: You can use just the amount you want, be it a little “hello world” program or a large-scale web system. And you are only charged on-demand.
- The resource must have a standardized interface: The interface should be as universal as the 110v power-plug or USB cables so that the cross-platform usage and the complexity overhead is minimized.
The true power of the cloud is its ability to decouple the computational resource and device’'s physical size. With the help of the cloud, your phone can harness the computational power of a supercomputer, and it certainly can go beyond that. This is the power that makes the mobile device AI really practical.