Killer Robots: A Utilitarian Perspective

Argument for AWS

The utilitarian argument for autonomous vehicles, weapons, and systems in general is that they supply on average the most benefit (utility) with least cost among possible options (e.g. not implement autonomous vehicles at all). The utilitarian perspective would then mandate that autonomous systems be pursued since they results in the maximum net benefit.

In the context of autonomous weapon systems (AWS), utilitarians would argue for their development.

The argument for autonomous weapons systems follows from the Trolley problem and other versions of it. In the Trolley problem, we might have a train on course to hit 5 people, but a lever could be pulled to save those people but result in the death of another person. The utilitarian would argue to pull the lever since it saves the most lives.

To understand this position more fundamentally we propose the following point. We posit that, barring extreme cases, the value of any two people is equivalent, or at least cannot be objectively differentiated and therefore are effectively equivalent. This would mean that the value of multiple lives is strictly larger than that of one life; five lives is greater than one.

So if our objective is to maximize the expected net benefit from some set of choices then we should select the choice that does just that. In the case of autonomous systems that seems to favor their implementation.

Autonomous weapon systems will be far more effective than human warfare methods. First off, they won’t require human input once deployed. This present several benefits. There will be cost savings since fewer people will be needed to operate weapons. Autonomous systems can be more accurate and efficient in executing their orders than humans might be, which means fewer casualties will occur and less unintended damage will be dealt on average. Furthermore since AWSs don’t require human control, there could be fewer deaths and injuries of military personnel.

In comparison to current methods, AWSs present improved benefits from financial and safety perspectives. If more dollars are saved and fewer innocent lives are lost, then the net benefit of AWSs outweighs other options and therefore they should be developed.

Discussion of responsibility

It is always possible that an AWS acts in a manner that people might not. Further, such an act could result in a significant number of casualties. Obviously, this is not ideal, but the utilitarian viewpoint still holds. In the long run, net benefit is maximized as more lives and resources are saved at a cost of rare incidents where there may be casualties and unintended damages.