Ethical Robots

This summer I came across a very interesting set of papers written by Ronald Arkin at the Georgia Institute of Technology. These papers outlined the foundation of an ethical control and reasoning system for autonomous warfare platforms and would reasonably include both ground robotic systems and Unmanned Aeriel Vehicles (UAV). This software was designed around the U.S. military's Laws of War and Rules of Engagement and funded in part by the U.S. Army.

Meanwhile, in the news where reports of lethal aerial drone attacks on targets in Pakistan. Although these attacks where controlled and authorized by human controllers, it is not far outside the scope of the technology to enable the autonomous platform to make its own lethal decisions. Foster-Miller, a leading maker of warfare robots, has introduced a heavily armed version of their TALON platform for use in combat. YouTube videos show this robot equipped with a machine-gun and engaging dummy targets. See here. Again, these devices a currently controlled exclusively by a human operator and no artificial intelligence is currently in charge of fire control. But it shows that the hardware exists and the only existing challenge is one of software.

The current direction of research is toward increasingly autonomous systems, including independent fire control. This technology is no so far off. Therefore, ethical software on these machines is needed to prevent them from making unethical or illegal decisions.

The popular fear surrounding this technology is the "Terminator" or "Matrix" scenario in which vast numbers of sentient military robots turn on their human creators and either enslave or massacre the entire human-race. This concept it far from reality at this point as robots a far from conscious and still need human intervention to remain operation (such as recharging the batteries.) But this fear is understandable. The thought of unfeeling machines armed with weapons and unable to differentiate between a soldier armed with a AK-47 and young girl holding an ice cream cone is unnerving. I can safely say I wouldn't want to be within range of one of these robots in constant fear that a single glitch could result in the robot firing armor piercing rounds at anything that moves.

Ultimately progress toward this goal will continue and one day, soon, we will hear about completely autonomous aircraft and ground robots on the fighting on the world's battle fronts. For this to be a reality these machines must be embedded with ethical logic and control. I for one hope that this will bring about a new era of warfare where machines, not men, fight and military superiority is determined by who has the best AI and strategy; much like a large, expensive video game. But I'm afraid this vision may be overly naive. Regardless of the outcome, it is essential that we imbibe these machines with at least a trace of humanity before we equip them with weaponry.