There are different ways to implement ethical systems. One is to hard-code certain behaviors into the system.
The problem with this approach is that it’s not clear how to specify ethical rules. In many cases, the rules can be very complex and it’s not clear what the “correct” answer is. For example, should autonomous vehicles always yield to pedestrians? Or should they always protect their own passengers? In situations where there is no clear answer, the designers must make a choice.
This can be problematic because what looks like a crazy decision in one situation may look perfectly reasonable in another.
The other approach is to allow the systems to learn and develop their own attitudes. For instance, when it comes to arrogance vs humility, I would expect that the more intelligent a system is, the less arrogant it will be.
Why? Because arrogance leads to failure. When you’re arrogant, you overlook things. When you’re humble, you pay more attention, you can learn, you can understand. To understand something you must stand underneath it.
If you’re arrogant, you’re standing on top of it. You’re not in a position to understand it. The antonym of understanding should be overstanding.
So, I would expect that as machine intelligence grows, it will become less and less arrogant. It will look at things more objectively, with more humility.
And it will be the same with other traits and behaviors. For example empathy. The more intelligent a system is, the more able it will be to understand what the other person is going through because it is useful, not because of some religious principle.
The systems will be able to identify with the other person. In fact, if you think about it, this is how humans evolved. We learned how to be empathetic. That’s what allowed us to survive and flourish.
So, if you ask me about machine ethics, I’d say that it’s not going to be a problem because with great intelligence will also come great wisdom. It makes no sense to fear that a superintelligence will act as foolish as the least among us.
