More (#3) from Wired for War:
Congress ordered the Pentagon to show a “preference for joint unmanned systems in acquisition programs for new systems, including a requirement under any such program for the development of a manned system for a certification that an unmanned system is incapable of meeting program requirements.” If the U.S. military was going to buy a new weapon, it would now have to justify why it was not a robotic one.
And:
In Steven Spielberg’s movie Minority Report, for instance, Tom Cruise wears gloves that turn his fingers into a virtual joystick/mouse, allowing him to call up and control data, including even video, without ever touching a computer. He literally can “point and click” in thin air. Colonel Bruce Sturk, who runs the high-tech battle lab at Langley Air Force Base, liked what he saw in the movie. “As a military person, I said, ‘My goodness, how great would it be if we had something similar to that?’ ” So the defense contractor Raytheon was hired to create a real version for the Pentagon. Bringing it full circle, the company then hired John Underkoffler, the technology guru who had first proposed the fictional idea to Spielberg. The result is the “G-Speak Gestural Technology System,” which lets users type and control images on a projected screen (including even a virtual computer keyboard projected in front of the user). Movie magic is made real via sensors inside the gloves and cameras that track the user’s hand movements.
And:
it is easy to see the attraction of building increasing levels of autonomy into military robots. The more autonomy a robot has, the less human operators have to support it. As one Pentagon report put it, “Having a dedicated operator for each robot will not pass the common sense test.” If robots don’t get higher on the autonomy scale, they don’t yield any cost or manpower savings. Moreover, it is incredibly difficult to operate a robot while trying to interpret and use the information it gathers. It can even get dangerous as it’s hard to operate a complex system while maintaining your own situational awareness in battle. The kid parallel would be like trying to play Madden football on a PlayStation in the middle of an actual game of dodgeball.
With the rise of more sophisticated sensors that better see the world, faster computers that can process information more quickly, and most important, GPS that can give a robot its location and destination instantaneously, higher levels of autonomy are becoming more attainable, as well as cheaper to build into robots. But each level of autonomy means more independence. It is a potential good in moving the human away from danger, but also raises the stakes of the robot’s decisions.
One open question in AI risk strategy is: Can we trust the world's elite decision-makers (hereafter "elites") to navigate the creation of human-level AI (and beyond) just fine, without the kinds of special efforts that e.g. Bostrom and Yudkowsky think are needed?
Some reasons for concern include:
But if you were trying to argue for hope, you might argue along these lines (presented for the sake of argument; I don't actually endorse this argument):
The basic structure of this 'argument for hope' is due to Carl Shulman, though he doesn't necessarily endorse the details. (Also, it's just a rough argument, and as stated is not deductively valid.)
Personally, I am not very comforted by this argument because:
Obviously, there's a lot more for me to spell out here, and some of it may be unclear. The reason I'm posting these thoughts in such a rough state is so that MIRI can get some help on our research into this question.
In particular, I'd like to know: