I suspect the default state of AI would be psychopathy. As I imagine it, an AI with goals, but not morals or empathy would behave exactly like a psychopath. Maybe psychopathy is the default state and empathy and morals are just evolutionary add-ons t encourage cooperation.
You still end up with a psychopath with moral goals. The rules leave room for manipulation, exploiting weakness, feigning emotion without understanding, lying, etc.
I think there are still more great sci-fi stories to be extrapolated from Asimov's laws.
I don't, and I suspect the opposite may be true. It's easy to approach this from the perspective of Star Trek or early AI thinking, but that's flawed. We don't understand consciousness at present but we do understand a little bit about cognition and emotion, and we can infer what we see based on examples. It's tempting to think that emotion is the height of human consciousness (the classic Star Trek fallacy), but this is plainly not true. Animals much less mentally adept than humans experience emotion and feelings. Indeed, it seems quite likely that such responses are easier to bootstrap than self-awareness and fully developed perception of the world.
Personally I think we're more likely to see emotional robots before we see truly human level AI. There is still reason for caution of course because emotion does not rule out psychopathy. There is certainly a danger that malevolent or even merely competitive AIs could play the emotion game better than humans, and gain the upper hand (meaning: wealth, control, power, etc.) Consider how much emotion affects the entertainment industry, and the sheer amount of money involved (from TV and films to music to books to comics to games and even to restaurants and food, not to mention pornography and prostitution). Consider how much emotion plays into politics.
Overall I think Blade Runner and A.I. had it more right than Star Trek or Star Wars. It'll be interesting to see how it all plays out of course.
