You cannot teach HUMAN right and wrong to a machine. A machine can only know what is right or wrong in its logic and with deep learning, it will always follow what it will consider the majority solution among the data it is considering. These two limits (limited data and majority statistics) make such machines extremely dangerous because their logic will be to eliminate the minorities that reduce the potential of the majority and Google does not have access to some of the biggest data fields, and first of all China. Alibaba as for that is several times more important than Google, especially for daily individual data: actually, Amazon is better than Google as for that.