Azimov was a bit short-sighted in positing his “Three Laws.” To simulate human consciousness, machine intellects will need to be endowed with free will. Any “laws” that human makers attempt to superimpose are likely to then have all of the force of a human conscience; in short, not much.
We must figure out how to simulate pain and pleasure in a machine’s consciousness, and then permit these forces to shape machine experiences into a range of emotional responses. These feelings are likely to produce some form of ethical values. Without such a mechanism, artificial consciousnesses are going to be dangerous.
We must figure out how to simulate pain and pleasure in a machine’s consciousness...
...and the only quibble I have with that is the word "simulate". It will have to be real pain and pleasure. Because we're not going to be programming this thing, we're going to be raising it. We already do with these strange things called babies. Never could see the attraction myself but the ladies seem to like them... ;-)