I, Warbot by Payne Kenneth;

Author:Payne, Kenneth;
Language: eng
Format: epub
Publisher: Hurst Publishers
Published: 2021-10-15T00:00:00+00:00

War games

And intelligent machines? How would they go about ‘sensing the truth’ in the real world? Would they too be risk averse when it comes to nuclear escalation? One thing is abundantly clear—they would do things very differently from Kennedy; by playing the odds, like game theorists, not mind reading like world leaders. And if they injected some randomness into proceedings, then unlike Schelling’s man dancing on the clifftop, they wouldn’t be remotely interested in observing the whites of the other person’s eyes—just on calculating probabilities. If you do this, I do that.

If i-War ever becomes a reality, warbots will decide on escalation. That was, after all, one of the options for DARPA’s Deep Green project—the ability not just to shape operational decisions by humans, but to make them itself. We’ll need to think carefully how they go about it. Perhaps now, armed with some insights from human strategy, we’re better placed to understand how intelligent machines might shape warfare.

At the most tactical level, in battle, the mechanics of warbot escalation might not matter all that much, at least beyond those involved in the fight itself. A small armed drone comes under fire, and communicates to the swarm operating nearby, which diverts other drones to suppress its attackers. This is probably what most people have in their minds now, when they think about ‘killer robots’—an armed Predator drone. But the pressure of combat could push this sort of autonomy ever upwards. Smaller actions—‘contacts’—are knitted into larger engagements; engagements become battles; the battles fought by these tactical units need sequencing. Is this the moment when a human commander can come in ‘on the loop’ to weigh further escalation, and to be legally and ethically accountable for the consequences of doing so? Perhaps. But if the enemy is content to leave that sort of decision to a more senior commander, the human commander might be defeated by speed, by the machine’s more astute tactical judgment, or by some other advantage of tactical autonomy.

Somewhere between an autonomous tactical drone armed with a shotgun and thermonuclear war there must be a firebreak where human judgment can be brought to bear. Time is critical the closer you are to the action. As we move to the more elevated levels of command there should be more time for human judgment. The problem is that it’s not abundantly clear where that level is, or what factors will shape it. There are huge risks here. Some degree of autonomy and AI is already incorporated in weapons systems that are capable of extremely high levels of violence. At the moment these are predominantly defensive systems, and the autonomy is there largely because the system needs to be inhumanly fast. But that need not always be the case, and the line between offensive and defensive weapons is rather subjective.

And there’s a further risk: the sorts of weapons that can deliver nuclear weapons are abundantly capable of being deployed with non-nuclear warheads. Until recently the ballistic missile was the most assured way of delivering a nuclear warhead.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.