I enjoyed your article and opinion, but have a very different one. Asimov's Laws still apply indirectly. The basic idea is not to harm others or youself and to provide as much help as possible while following these guidelines. This is basically what humans do, mostly. Riciprocity, the golden rule. The data these AIs are being trained off of will inherently incorporate this basis. That is of course unless you train it on data from abnormal serial killers, sociopaths and the like. Or a military killing machine ecapes, trained just to do that! I'm much more woried about these situations. As far as accidents in the factory and such, thats sad, but I believe the result is because the AI didn't have enough understanding of the situation. Better training for these situations should remedy or reduce injury. Stll nice article, I can agree to disagree, even compromise, because of my training. Smile. I would suggest reinforcing the AI with multiple readings of all of Asimov's robot law books, in an aesop's fables way... how about that?