Philip Rosedale has a pretty provocative Medium post arguing against artificial intelligence as the future of industry (something generally assumed to be true in Silicon Valley), and instead, argues that robots, controlled by humans via VR, is a better path ahead:
VR headsets that also have hand controllers are now widely available and are being used mostly for computer gaming. But these headsets could be re-used to create a workstation at home allowing you to slip on a VR headset and be ‘teleported’ into the eyes and body of a robot: Wherever you look, the robot looks. When you move your hands, the robot moves it’s hands. So it is exactly like being somewhere else, except you are ‘in’ the body of the robot. But of course that robot could be much physically stronger than you, or it could be made very large, or very small, or whatever else the job demands. Plus you would be able to talk and listen (through the robot), so you could interact with people or other robots just like in real life. So, for example, a firefighter could rush into a burning house as a robot, and be unaffected by heat or smoke but still able to do his job and save people using the same instincts and training he’d learned over his career.
Philip's argument suggests a solution for countless workers in industries afraid that the coming of AI will take away their jobs. While most of the technorati assume AI-powered self-driving cars are inevitable soon, for instance, Philip argues instead that there's a place for automotive vehicles powered by humans, that self-driving technology can't soon replace -- if it's done in VR.
Some industries might in the long term have people replaced partially or entirely by AI, but this is going to take a long time, and for some might not happen at all. Remote-operated robots, by comparison, could take on most of these jobs today. Consider the case of Uber: many people anticipate the day when Uber vehicles will be self driving. But if you unpack each of the different steps required to make that possible, from identifying the specific rider waving at you in a crowd, to finding a place to safely pick up and drop off, to checking that the vehicle itself is clean and ready for the next pick-up, you quickly amass a long list of features that will be taxing or impossible for near-term AI systems. The practical way to achieve a ‘self-driving’ car will likely still need to switch over frequently to remote human assistance — which will probably be done with remote VR operators.
It's important to point out this is already happening: Military drone operators basically operate in a VR environment, and make literal life or death decisions we definitely don't want to devolve to AIs. At the same time, this points out one dark side to Philip's vision: Drone operators suffer as much PTSD as real life military pilots.