Interesting thread, thank you for starting it.
Eventually, yes. At least that is the plan for the US military.
Have they given any idea of the sort of direction they want to take that? After all, military technology has been improving since somebody first decided to tie a stone onto a stick, but we've consistently failed to find anything that replaces the infantry soldier and what he can carry. The well-worn phrase is that only infantry can hold ground - riflemen can get into small spaces (such as river-beds and buildings), move through difficult terrain, use cover and concealment and move around fast on a tactical scale. This is why the introduction of cavalry, artillery, armour and air power all failed to take away the need for something basically the same size and agility as an infanteer with more-or-less light equipment. Infantry soldiers are also able to operate completely autonomously, make creative decisions about unpredictable situations and deal with human beings - think of the sort of reassurance patrols usually conducted with berets and slung rifles. By the time you've made a machine that can do all of that, I think you've basically rebuilt a human being, and probably one complicated enough that you can no longer say that they are more disposable than a person. There's certainly room for technology to cut down the infantry's workload: I can imagine recce patrols disappearing as drones and satellites get better, for example. But it's not just that we don't have the technology at the moment to replicate the infantry soldier: I can't at this stage imagine what such a replacement would look like, with the best technology in the world.
There are then other thorny issues:
I suspect that no one will ever be really comfortable with letting machines make actual life and death decisions, so killing will always be the prerogative of men...who will most likely send boys.
Ethics become very different when you're dealing in algorithms. At present, there's a degree of human error. Every now and again, a civilian laying pipes will be mistaken for an insurgent with an RPG and shot. It's a very different thing when that's a 19-year-old making a reasonable but wrong judgement call, versus when that's a pre-programmed system that will act the same in all situations. Accepting a degree of collateral damage is one matter, but having machines wage war would require coding in a quantifiable amount of it, and situations in which it would occur. The same problem occurs with self-driving cars. You have to program the machine to be able to decide between swerving into traffic and swerving into pedestrians, and that means you need moral answers that nobody really has.
If anything, military training shows up the basic problems in using a machine that can't learn as well as a human being. Tactical training is really an exercise in programming: if you think of the Sandhurst/Brecon/West Point style platoon battle drill, teaching that is a matter of getting people to automatically go through a pre-written process of thought and action. This is at once the best way to teach it and, as we all recognise, totally inadequate: we recognise that officers and NCOs will always be a bit unfinished until they have combat experience to give them the sort of instincts that can't be made to fit such an algorithm. To replicate that, you'd need to have a machine whose brain worked quite a lot like a human one, and would end up with a unique set of thought processes - in other words, a personality. Once you've made a machine like that, I'm not convinced you can say that it's still a machine.
Have you actually participated in a modern battle? I have, and I can say with complete certainty that this statement is not accurate. The technology we are adding to our military has drastically changed the way we fight, especially when it comes to asymmetrical warfare. The most easily recognizable difference is the number of soldiers required to carry out an operation or hold a given position. The addition of drones and other "gadgets" has greatly reduced the actual number of soldiers we need to deploy. I remember our commanding officer talking about this once and he said basically what we needed a company-level element to do 20 years ago can now be easily accomplished by a platoon-level element. We are entering an age of warfare where numbers are truly becoming irrelevant. We live in an era now where the biggest army isn't necessarily the strongest, it's the army that has the better force multipliers and the better trained soldiers who can use those force multipliers effectively.
That's been the American doctrine, more or less, since about 1944. Recently, though, we're starting to see a bit of a push-back against it. The Royal Navy, for example, spent the past few decades building better and better ships, and boasting that a new frigate can do the job of six of its predecessors, which is entirely true - except the new frigate can only be in one place at once. You can think of similar examples with land forces - if you want to control access into a village with three entrances, you need three groups of (at least two) people. You then need a degree of redundancy: if Smith falls over and breaks his leg, that's manageable if his whole platoon are holding the village, but much worse if there are only six of them.
I agree that the trend is going that way, but I don't agree that everything is moving towards zero, or that we've simply crossed out 'company' in 'company-level task' and replaced it with 'platoon'.
Nah, it was fifth largest, so I wasn't too far off:
I can't remember who, but as one American general put it, the 2003 invasion turned the Iraqi army from the fifth-largest in the world to the second-largest in Iraq.