I think the next five years will be software extracting shocking performance from existing machines. The next ten will be designing machines no human could possibly pilot.
I'm radicalized about the kind of vehicles we should build. We should make much stranger machines.
I just watched a wheel loader do acrobatics on its wheels, controlled by software. A heavy machine throwing itself around.
I think this is a preview of a renaissance in control systems. We're about to discover how much performance is hiding in hardware we already have.
This doesn't require an "AI pilot" making decisions the way a person does. The breakthrough is control and optimization software that understands its machine.
Software can coordinate motions no person could keep straight, and use hardware closer to its true limits.