How to design Artificial Intelligence for the people's good

In the famous sci-fi book (and movie), the "precrime" police is able to stop murders before they happen thanks to humans who predict the future. The concept is fascinating and is an interesting example of a possible use of Artificial intelligence. But we would leave the decision and control fully to the machines which we must not let happen in the real world. People have the right to own their future and no machine can be deciding on their behalf. When I read articles such as the recent one in La Stampa by Lorenzo Longhitano that a computer can predict what a chance of curing a sick person is and whether it is reasonable to admit her or him to the hospital, I think we need to draw a clear line.