Can We Pre Program Machines To Be Ethical?

Some solutions put forward are-

• Specifying rules to determine correct decisions taken by AI systems.
• Maintaining a degree of human input which would kick in when an ethical decision had to be made.

However, the problem with specifying what a machine should do is that AI encounters unexpected and unpredictable situations a lot of times that it's too hard to specify what an AI should do in advance. The problem with the second solution is that some decisions need to be made in a split second and in some circumstances it won't possible to get human input.