I think the search for a suitable commander for the robot armies is a silly one. Asking "Who will command robots?" is like asking "who will run software?" Ever

I think the search for a suitable commander for the robot armies is a silly one. Asking "Who will command robots?" is like asking "who will run software?" Everyone will. What will they do? Whatever their creators and owners want them to. There are already a lot of robots in the world, just as there is already a lot of software. Indeed, Hollywood likes to depict robots by starting with a human mind and warping it, but a better intuition is formed by remembering that robots are software with physical outputs. They aren't going to spontaneously develop new abilities unless they have been created to. Some people are inclined to develop software carelessly. "Move fast and break things" is a sensible motto for someone at the center of building social media from nothing. That's only one engineering culture, though. Visit labs in heavy industrial R&D, in academia, in defense, listen to open source developers' chatter, and you will see plenty of people write books on whether a single line of code should say this or that. Cheer up. It isn't all maximum velocity startups. :) Software - and robots as instantiations of software - is not good and it's not evil and it's not careless and it's not careful. It can be any of these, because it is a reflection of its creators. Like all machines, it amplifies their will. I very much expect robotics to follow the same path that automation before it always has. In the beginning, industrial machinery was dangerous to the workers who used it, being first generation tech designed only to get a new job done. As the decades wore on, the machines became safer, quieter, smarter, more predictable. Best practices and required safety protections developed, and now industrial machinery maiming someone is a scandal. Machines of warfare were once indiscriminate. The buzz bombs and incendiary balloons of the second world war would be considered terrorist weapons today. The weapons that kill indiscriminately after a conflict ends - the land mines, the chemical weapons - banned now. Sure, they're still used by those who don't care about their moral reputation, but it's the sort of thing that makes the international community talk about putting together a coalition to overthrow your government. Nobody does it lightly. Would the militaries of the world make the same mistake with robots - not retaining control of every kill? Putting an experimental-ai long-lived robot behind a weapon where it might kill for all the wrong reasons after a conflict is over? I seriously doubt it. Too many people would see such a thing as a land mine, and rightly so. A military that didn't care about their moral reputation might do that, but that doesn't make responsibly developed military robots bad. Machines reflect their creators. I very much expect that whatever mistakes the militaries of the world make with robots will be new ones, not easily foreseen. And that as we get better at the technology, we'll correct those too, and ultimately - just like our industrial engineering, just like our software in general - they will be as safe, and as dangerous, and as predictable, and as smart, and as useful, and as directly controlled by humans, as we as a society want them to be. Mature technology doesn't look like Facebook. Never has, never will. It looks like Apache - something well-considered, controllable and safe, something which does a job so reasonably and so well that it's boring to everyone but the geeks who work on it. :) Robot armies will be controlled by everyone, and they will be as unremarkable as the robots I control now: my quad, my 3d printer, my washer and dryer, my car. Dangerous if misused or badly designed, but these things are are avoidable, and beyond that - safe, convenient, nice to have.