Bright lights, like this loon, are all part of the "man is not ready......." pseudo religious bullshit".
In fact, we will progress to artificial life and artificial intelligence in erratic steps - some large, some small - some hard, some easy.
Easy is logic, easy is memory and lookups, easy is speed - hence Watson as we start to climb the connectedness/co-relatedness/content addressable memory ladder. (Content addressable memory {CAI } is like a roll call in the Army - "Private Smith?" - "here"). A lot of the aspects of intelligence are ramifications of CAI, and other aspects of interconnectedness. Add in the speed and memory depth and more and more aspects of an AI emerge. As time goes, step by step, intelligence will emerge. It might be like an infant that needs to learn as we do, but at a far higher speed - zero to 25 years old in 5 minutes???. Experiential memories - can they be done at high speed, or must that clock take longer?.
The precise timing of these stages elude me, but I believe they will emerge with time.
As to whether or not this AI will be a malevolent killer, or one of altruistic aspect??? It seems to me that this will depend on how is is brought up.
(until an AI can reproduce sexually - no he/she). Can a growing AI be abused - mentally, as in children are abused?? I suspect that with no sexuality that there will be no casus abusus. That is not to say that ways to abuse a growing AI are impossible to find - they will emerge in time.
As these AIs emerge, how smart will they be? and IQ of 25 or one of 25,000,000?? This might bear some relationship with how these AIs treat mankind, as a student at man;s knee, or as something that looks down at man with an IQ of 100 and also sees bees and ants with with a group IQ of 25? and muses - what's the difference and thinks of other things...