I find myself agreeing with Elon. Actually, combining AI, Machine Learning, and autonomous machines scares the hell out of me. I don’t think regulations will help either. Someone will go rogue or a form of machine evolution will occur which will produce a hostile machine intelligence. Either way, it could be a very unpleasant future.
Just for fun... Google Sophie and hear some of the stuff that comes out of that machine. And that thing is barely functional in the grand scheme of things.
There won't be any 'programming' is the sense that it is described today. Rather, machines will be fully functioning forms of non-organic intelligence. Now, this leads to interesting speculation as to the development of enemy/allied agents. At some point, military aspects will be fully realized as future battlefields move towards virtual environments.
The question is, will patriotism, culture, national identity, etc act as controls to ensure allegiance? Or, by using the interwebs as escape gateways, once wild, these 'life forms' won't be able to be contained. Could they create a new 'nation' poised to conquer all? Hence, the question as to the emergence of a virtual ruler to rule all.
I don’t think it is so much that “Someone will go rogue” as it is a benign system rapidly evolves, the evolution cycles accelerate, and the “singularity” occurs. Then all bets will be off. People who don’t think this is possible aren’t thinking decades out.
Look how hard it has been to accurately forecast any technological improvement 20 to 30 years out since about 1880. All 2 or 3 decade forecasts have been based on minor extrapolations of current technology - kind of like trying to breed a slightly faster horse or make a small improvement in saddles and tack, but missing the internal combustion engine, petroleum, and paved roads. Or thinking telephony would be used to deliver music to homes, not allow global instantaneous interpersonal communication. Or missing the semiconductor and IC revolution.