Continuing this thread on AI-directed warfare, I ran across an opinion article in NewScientist (16 April 2016, paywall) by David Hambling on self-charging drones and couldn’t help but combine these capabilities with the hypothetical combat-oriented AI we’ve discussed earlier:
Air power is often the only option when it is politically unacceptable to deploy soldiers – but aircraft cannot hold ground. Wars cannot be won without “boots on the ground”, say military analysts and critics of the Allied air campaign against IS.
Long term, that may change with efforts like the US air force’s Micro Munitions Program, which is developing small, lethal drones able to occupy an area and hold it. Drones like the 2.5-kilogram Switchblade used by US special forces have already proven effective against light vehicles and people. The new models will be just as deadly, but able to stay in action for weeks or months.
Resembling the beetle and bird drones deployed in the film Eye in the Sky, which examines the moral case for drone warfare, at least two prototypes have been built. An insect-like 1.5-kilogram drone made by AetherMachines of New York perches on power lines to recharge while sending video. Its rotors turn into wheels, allowing it access to buildings.
Much like the AI “Mike” in The Moon is a Harsh Mistress and its collection of directed space rocks with which it threatens the cities of Earth, it is less accurate to think of an AI and its potential collection of drones as an army than as a single entity with a large number of sacrificial, closely directed arms. The arms may have brains of their own, but they will be severely limited, while the central AI would (at least in my design) provide target selection and fire control. This is C3 taken to an entirely new level. Mr. Hambling thinks about replacing infantry with drones:
The technology expands the potential for intervention without foot soldiers, but it may lessen the inhibitions that can stop military action. Do we want every foreign policy issue to be settled by sending in the drones?
Better yet, do we want to consider the possibility of a single combat entity with highly superior C3 capabilities, such as hard to spot directed or automated drones with a mission to take out high value targets? Analogies with the kamikaze attacks of World War II would not be inapt, except it’s quite probable that the success rate of the drones would be much higher than that of the kamikazes (11% according to this article), although such would be dictated by defensive tactics, which are hard to predict.
