With respect to my post: I was responding to his question "What's so bad about that?" which took replacement as a premise. Generally: I'd prefer an upgrade, bu

With respect to my post: I was responding to his question "What's so bad about that?" which took replacement as a premise. Generally: I'd prefer an upgrade, but I worry that if we achieve strong AI it will be an extremely difficult problem [1] to ensure it behaves nicely -- so replacement may be more likely than upgrading. If we focus on augmenting human intelligence and not on developing independent intelligence this may be less of a problem, but I don't know which research paths will be more fruitful. [1] http://wiki.lesswrong.com/wiki/Friendly_artificial_intelligence