Homepage Feature

Programming an ethical robot warrior

Audio: Play | Download

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Download MP3
Technology that thinks for itself is nothing new for fans of sci-fi movies like 2001: A Space Odyssey. In real life, people still have the final say over computers when it comes to making decisions, especially life-or-death decisions out on the battlefield. But some researchers are working to develop software that will help robots make moral and legal decisions on their own. We speak with Ronald Arkin, a professor of computer science at Georgia Tech. He’s just completed a three-year research project for the Army looking into the use of ethical battlefield robots.

• Ronald Arkin’s homepage
• BBC: Can battlefield robots take the place of soldiers?
• BBC: Robots of the future unveiled

Read the Transcript
This text below is a phonetic transcript of a radio story broadcast by PRI’s THE WORLD. It has been created on deadline by a contractor for PRI. The transcript is included here to facilitate internet searches for audio content. Please report any transcribing errors to theworld@pri.org. This transcript may not be in its final form, and it may be updated. Please be aware that the authoritative record of material distributed by PRI’s THE WORLD is the program audio.

KATY CLARK:  Technology plays a key role in any U.S. military operation, including the current one in Afghanistan.  But by and large machines deployed in the battlefield still rely on humans to make decisions for them.  Now, though, researchers are working to develop software that allows military robots to make their own decision.  Ronald Arkin is a professor of Computer Science at Georgia Tech.  He’s just completed a three year research project for the Army looking into the use of ethical battlefield robots.  We’ve all heard of unmanned aircraft being piloted by remote control from thousands of miles away.  Professor Arkin, how else are robots being used on the battlefield right now?

RONALD ARKIN:  Well Katy they’re being used in a variety of different ways.  There are certainly ground robots if you saw The Hurt Locker, for example, you’ve seen explosive ordnance disposal robots made by companies such as iRobot.  There are surface vehicles being developed by the Navy.  There are also undersea vehicles in addition to aerial vehicles.  So the unmanned spectrum for potential autonomy, not quite autonomous yet, goes from underwater, to the surface of the water, to the land and up into the air.

CLARK: But existing robots that you’re describing are still controlled by humans at least somewhere.  How could people be taken out of the loop?  Tell us a little bit about your work to get to this point.

ARKIN: Right, well it’s not exactly people being taken completely out of the loop.  In a military scenario there’s always someone in the loop.  We are now finding that autonomous systems, the decision-making is being pushed further and further forward to the so-called tip of the spear due to the ever-increasing tempo of the battlefield, the speed at which soldiers have to respond due to the increase in weapon capabilities.  For that particular reason, we find that autonomy, the ability to make decisions, even the ability to make a decision what to target and when to actually release the weaponry, is going to become more and more prevalent.

CLARK: So it all comes down to programming and you’re looking at moral, legal decisions being distilled into a computer algorithm?  If so, how does that actually work?

ARKIN: Right.  Well, it’s actually not hard to tell a robot to pull a trigger on a gun.  That’s the easy part.  The hard part is to tell the robot to pull a trigger on a gun that it has adequately and effectively discriminated its target, that it has made sure that the people that have been involved in the tasking of this system are completely responsible for its particular use; that it is acting with proportionality, in other words choosing the right weapons systems that match the particular situation at hand.  These are the sorts of things that at least I am trying to design into these systems.

CLARK: And I know that you’re also looking beyond just the moral and the ethical elements; you’re looking at putting emotions into these war-fighting robots too, aren’t you?

ARKIN: Where are emotions appropriate in the battlefield?  As I’ve written, anger and hatred and many of the emotions are actually ill-advised to build into these systems because they actually cloud the judgment of even human war fighters under those circumstances.  But there is a special set of what’s so-called “secondary emotions”, the moral emotions, which includes things like empathy and compassion and shame and guilt.

CLARK: Guilt.

ARKIN: Exactly guilt.  And guilt is the one that we have chosen.  Can these systems do better?  Not necessarily perfectly.  But if they can do better than humans can do in the battlefield, what that translates into, and this always sounds strange, is the saving of human life.  It results in the saving of non-combatant casualties and a reduction in non-combatant collateral damage.  That’s what this research is about.

CLARK: So what you’re saying basically is, we’re at the point now where we can build these ethical war fighting robots, but the question remains should we?

ARKIN: I would not quite go that far.  I would say that we have developed a proof of concept in our research that points the way to the potential creation of these systems.  The whole point is to lead to international discussions regarding the appropriate use of this technology.  We need to make decisions before it happens, proactively, not reactively.

CLARK: Ronald Arkin is a professor of Computer Science at Georgia Tech.  He’s also the author of Governing Lethal Behavior in Autonomous Robots.  Thank you so much.

ARKIN: It was my pleasure.  Thank you Katy.

Copyright ©2009 PRI’s THE WORLD. All rights reserved. No quotes from the materials contained herein may be used in any media without attribution to PRI’s THE WORLD. This transcript may not be reproduced, in whole or in part, without prior written permission. For further information, please email The World’s Permissions Coordinator at theworld@pri.org.