Depends on your morals, right? For example, a rational Egoist shouldn't have too much moral trouble with such a solution (of course, he may still reject it on practical grounds, as the idea of controlling a posthuman is questionable). In fact, the way we raise our kids and indoctrinate people to keep them from anti-social and criminal behaviour is a universally accepted and indeed necessary form of programming. So if you accept this, then there is no *moral* reason not to accept putting some basic moral rules into an AI.
> I suggest that this passage be amended to remove the advocation of mass mind
> control. Perhaps something like this:
> In the first case, we could make sure that the first such entities possess
> a thorough understanding of, and respect for, existing human moral codes.
In the first case, we should make sure that such entities are *us*.
In the first case, we should make sure that such entities are *us*.Creating separate SIs (from AI) is a BIG mistake. We can certainly make sure that the new entities thoroughly understand us, but that by no means guarantees their respect for our moral codes.
> That would be enough to ensure that they at least think the whole thing
> through before adopting some completely alien viewpoint. It is also the
> strongest measure that I can see either a moral or a practical basis for.
Morals are subjective, but the most practical (=rational) course of action would be not to create AIs with SI potential and work on uploading instead. Again, our prime directive should _always_ be survival. Survival is the prerequisite for _all_ other actions. Any philosophy that does not value personal survival [in an optimal state] above everything else is by definition irrational. Thus follows that transhumanism (with an immortalist element) is the best philosophy currently available to us.