– Bernd Debusmann is a Reuters columnist. The opinions expressed are his own –
They have no fear, they never tire, they are not upset when the soldier next to them gets blown to pieces. Their morale doesn’t suffer by having to do, again and again, the jobs known in the military as the Three Ds - dull, dirty and dangerous.
They are military robots and their rapidly increasing numbers and growing sophistication may herald the end of thousands of years of human monopoly on fighting war. “Science fiction is moving to the battlefield. The future is upon us,” as Brookings scholar Peter Singer put it to a conference of experts at the U.S. Army War College in Pennsylvania this month.
Singer just published Wired For War - the Robotics Revolution and Conflict in the 21st Century, a book that traces the rise of the machines and predicts that in future wars they will not only play greater roles in executing missions but also in planning them.
Numbers reflect the explosive growth of robotic systems. The U.S. forces that stormed into Iraq in 2003 had no robots on the ground. There were none in Afghanistan either. Now those two wars are fought with the help of an estimated 12,000 ground-based robots and 7,000 unmanned aerial vehicles (UAVs), the technical term for drone, or robotic aircraft.
Ground-based robots in Iraq have saved hundreds of lives in Iraq, defusing improvised explosive devices, which account for more than 40 percent of U.S. casualties. The first armed robot was deployed in Iraq in 2007 and it is as lethal as its acronym is long: Special Weapons Observation Remote Reconnaissance Direct Action System (SWORDS). Its mounted M249 machinegun can hit a target more than 3,000 feet away with pin-point precision.
From the air, the best-known UAV, the Predator, has killed dozens of insurgent leaders - as well as scores of civilians whose death has prompted protests both from Afghanistan and Pakistan.
The Predators are flown by operators sitting in front of television monitors in cubicles at Creech Air Force Base in Nevada, 8,000 miles from Afghanistan and Taliban sanctuaries on the Pakistani side of the border with Afghanistan. The cubicle pilots in Nevada run no physical risks whatever, a novelty for men engaged in war.
TECHNOLOGY RUNS AHEAD OF ETHICS
Reducing risk, and casualties, is at the heart of the drive for more and better robots. Ultimately, that means “fully autonomous engagement without human intervention,” according to an Army communication to robot designers. In other words, computer programs, not a remote human operator, would decide when to open fire. What worries some experts is that technology is running ahead of deliberations of ethical and legal questions.
Robotics research and development in the U.S. received a big push from Congress in 2001, when it set two ambitious goals: by 2010, a third of the country’s long-range attack aircraft should be unmanned; and by 2015 one third of America’s ground combat vehicles. Neither goal is likely to be met but the deadline pushed non-technological considerations to the sidelines.
A recent study prepared for the Office of Naval Research by a team from the California Polytechnic State University said that robot ethics had not received the attention it deserved because of a “rush to market” mentality and the “common misconception” that robots will do only what they have been programmed to do.
“Unfortunately, such a belief is sorely outdated, harking back to the time when computers were simpler and their programs could be written and understood by a single person,” the study says. “Now programs with millions of lines of code are written by teams of programmers, none of whom knows the entire program; hence, no individual can predict the effect of a given command with absolute certainty since portions of programs may interact in unexpected, untested ways.”
That’s what might have happened during an exercise in South Africa in 2007, when a robot anti-aircraft gun sprayed hundreds of rounds of cannon shell around its position, killing nine soldiers and injuring 14.
Beyond isolated accidents, there are deeper problems that have yet to be solved. How do you get a robot to tell an insurgent from an innocent? Can you program the Laws of War and the Rules of Engagement into a robot? Can you imbue a robot with his country’s culture? If something goes wrong, resulting in the death of civilians, who will be held responsible?
The robot’s manufacturer? The designers? Software programmers? The commanding officer in whose unit the robot operates? Or the U.S. president who in some cases authorizes attacks? (Barack Obama has given the green light to a string of Predator strikes into Pakistan).
While the United States has deployed more military robots - on land, in the air and at sea - than any other country, it is not alone in building them. More than 40 countries, including potential adversaries such as China, are working on robotics technology. Which leaves one to wonder how the ability to send large numbers of robots, and fewer soldiers, to war will affect political decisions on force versus diplomacy.
You need to be an optimist to think that political leaders will opt for negotiation over war once combat casualties come home not in flag-decked coffins but in packing crates destined for the robot repair shop.
Robots can aid and assist in combat, therefore relieving or helping certain aspects of a soldiers duty. But, robotics technology will never get to a point to replace soldiers. Mankind as a whole has always used tools and weapons as a means of fighting, and robots are just another level of weaponry that we are using. Robots are programmed to do what the programmer wants it to do. A human programmed it to do it; the concept of AI is of huge debate, but since true AI is near impossible to achieve (at least by current technology), and even pseudo-AI isn't really AI (still only programmed by humans) I don't see any robot replacing a good soldier with the instinct and the intuition to fight in real combat, anytime soon.
Trackback
80 comments so far
Previous | 4 | 3 | 2 | 1 | Next
[...] It is just a Sci-Fi channel hoax that military grade robots are right around the corner, well no it isnt. Apparently 12-thousand are working in the Middle East right [...]
Absolutely fabulous. We should build more of these and keep them flying over Afghanistan and Palestine at all times. The idea about using them to kill Somalia pirates is even better. Keep up the good technology!
Designing a robot that can decide who and when to kill is dangerous. I work on computers for a living, and they often fail for many reasons. They have hardware failures and software bugs. They may be sensitive to heat or cold.
How can a robot tell who the “enemy” really is? If a robot meant for a war in another country were to get loose in America, how would it know the people it encounters are those it is suppose to protect?
This idea sounds like a bad sci fi movie. I wonder how long it will be before someone tries to infect battlefield robots with a computer virus?
There are many weapons of war that have been banned by civilized countries. This should be one of them.
yawn…. looks like SkyNet will be taking over soon…
Ethics - Will you accept fighting this unit in the battlefield?
yes = it is acceptable to make,
no = it is unacceptable to make.
For this question you need to consider yourself as the target.
Capabilities - How difficult would this be to block/kill for a human
fighting the unit?
What is the intended target and what “style” of combat employed?
Communications - command->response time-delays?
How susceptible to communications takeover/blocking is the unit?
If it has a human controller then it is a variation of a vehicle
transport and being representative(game piece) of the controller
Independant operation ? - this requires a level of programming
far higher and possibly in excess of a single processor unit.
(Multiple Independant Dedicated Processors with internal network)
The Ethics of this system is a reverse engineering debate as well
What processor finally pushes the fuzzy value for “shoot to kill”
into the “action zone” where it is permitted to go beyond target
recognition into actively killing the target?
How much detail is required?
personally I see this as not a single program but a stack result.
there are certain combinations of the Internet TCP/UDP/IP4/ICMP2
protocol stack results that *never* map due to non-occurence.
Development Security? - Hack the Development for self protection
best network security is no network (isolate the development)
When the prototype unit behaves as required, throw real humans
and *mandate* they attempt to break the units behaviour
Isaac Asimov had the robot psychology simplified,
Robert Heinlein went towards part of human psychology
I personally consider the situation as an asymetric problem.
any solution includes the person creating the solution as part.
Ive deliberately skipped a lot of details due to space.
I personally see combat requiring mixed forces at the non-ground
troop levels.
I also see “ground troops” having negative contacts with AI troop
movements when they meet in most situations entirely by training
unless there is some kind of standard recognition, but this will
require long-range sensors on the automated units.
my personal opinion on the whole matter is an automated army being
easily susceptible to mis-use of the command structure.
(all modern military suffer from this susceptibility imho)
I’m not learned of modern strategy and tactics outside video games
This requires further debate and a definite clear strategy as
any forward production or front-line equipment needs to improve
the battlefield for the military using it.
Existing technologies have expanded on or integrated with existing
military structures, future successes will be either following
a similar path (becoming integrated) or need to deliver
an “out of the ordinary” military capability that is not one-time
usable and is excessively effective.
a “capture” instead of “kill” mentality may be better serving,
“capture” of enemy units for re-programming or re-use as parts.
instead of flat destruction.
a more difficult strategy but also providing a higher reward.
capture of combatants and civilians and post-capture sorting
would also work to eliminate the collateral damage costs.
Moral dilemmas aside, the biggest problem that see is one of Horror. What is the point of a war? To force another person or political body to abide by your rules — to do what you want them to do — or die. As Patton said “The object of war is not to die for your country but to make the other bastard die for his.” Once you take the dying out of it, for one or both parties, you take the Horror out of what you are doing — it becomes just another video game it becomes… entertainment. For the side controlling the robots at least. Who cares if a robot gets destroyed or has arms or legs amputated? We can just put it back together again or build a new one and send it out again. Without the destruction and killing of ones soldiers it becomes easy for political bodies to declare war on each other. Why consider alternatives to war — why peace? War is easier. No one dies on our side. The enemy still ends up dead. Win - Win, right?
Well when one looks at how the partisan politics in DC and most state, county and city government “operates”, seems we already have a study group at hand.
The drones and droids in place in our “hallowed halls” and elected offices are a classic study group for mindless robots following mindless partisan and special interests and “$-sign” “programing”?
Coming from IT background, I can predict countries making robot will be most affected by its misuse. Hacking robot will be as easy hacking a computer. Believe me some one must be already waiting for USA to make such robot so they can turn them against USA.
Robots taking over the world, been watching a few too many Hollywood movies.
The reality is the current state of the art robots are not that dissimilar to remote controlled toy cars. So those ethics questions should be directed at the military commanders and not over hyped robots
And complex systems aren’t new to the world, thousands of people trust their lives to them every day. How many millions of lines of code keep a 737 in the air?
May I offer some measure of realism here?
So far I haven’t seen the head of Osama Bin Laden on a plate…did any of you? How does bringing down whole mountains with uber-explosives accomplish making sure someone is a) Dead, and most importantly b) Prove it to everybody beyond a shadow of a doubt…?
If investing countless millions of dollars in high end military gadgetry was genuinely useful to win wars, then the US should have obliterated any opposition in Iraq, Iran, Palestine and Afghanistan and anywhere else by now - with no casualties and inflicting the least amount of damage to infrastructure and civilian populations.
That is not the case. Guerrillas in Iraq and Afghanistan have proven to be resilient despite of their limited access to the type of high level of the american technology.
Seems like guerrilla men are a hands-on type of people who have vastly higher amounts of realism in their heads and use well whatever technology they put their hands on.
Any american military leaders who foolishly buy ultra-expensive gadgetry thinking they’re fighting the Klingons are sadly mistaken or are getting a cut of the profits. At this point in time, the line between private and the US Army is so blurred, you’d think they’re one and the same, except for whom they’re intended to serve.
The russian defence contractors are selling cheap inflatable balloons that do the job while the american military is constantly buying billion-dollar *stuff* that doesn’t necessarily help the men on the ground - who are the ONLY way to secure territory and assure victory.
A propos, just recently I had a garage sale to get rid of all the kitchen gadgets that were supposed to help me chop, slice and cook food without getting my hands dirty. Since purchased and used once, they remained inside a cabinet taking up precious room until there was no more room and my wallet left emptier. I should be less willing to trust the salesmen who claim their latest electrical onion chopper will simplify my life, especially when the gadget can’t be put into the dishwasher.
Having a knife and a cutting board is far easier, after all.
I think too many defence contractors are meddling in the affairs of nations far beyond acceptable and it’s about time to cut their meddling down to manageable size. I know the GDP of certain whole nations owes a lot to the american military buying their stuff but there are other products that could be created instead of killing machines and that would be just as profitable. It depends on what is in the heart of the manufacturers, isn’t it?
One should not forget about all the nuclear weapons (or future new weapon technology likely even more devastating than the ones we currently have - don’t forget about biological warfare technologies) that will always be present and be a factor in any future war, robots included or not. Will humankind ever allow robots to make the decision to press the nuclear button (or activate some super virus escape mechanism), and lead to the destruction of Earth and/or humankind (if we haven’t colonized other planets yet)? Don’t say no yet, as robots may well take over in some way at some point. How can one prevent any of these wars of robots as envisioned by other posters from escalating into an all-out war, including using, of course, all the weapon technology at humankind’s disposal (or at robots’ disposal – if they had already taken over at that point)? Any of the warring parties may not even need to have the means to deliver such weapons over to the other party’s territory – all they need to do is make sure that it explodes (or in the case of a super virus, simply allow it to escape)– anywhere – and humankind will cease to exist. There seems to be a good likelihood that self-replicating intelligent robots will still exist in the universe after all humankind has been destroyed, as robots are more likely to survive the super virus, if not nuclear weapons. Will robots even care about whether humankind survives or not?
“Two words, or rather the name of one author: Isaac Asimov.”
Asimov would be sickened by modern military robots which make a mockery of his three laws. Actually, it’s a good bet all the geniuses who’s work is being corrupted to create these moral obscenities would be. My favorite part of the article is when it mentions increased computer involvement in planning as well as executing warfare. Add a little AI and you’ll have Skynet for real. Far fetched? Not really, since AI is the only lacking element. And you know the military will be stupid enough to add it, as soon as they see a possible military advantage. Only a matter of time, really.
While I agree with you that the preservation of our soldiers lives should be a high priority, we must not overlook the possibility that advances in robotics and artificial intelligence will lead to semi-sentient killing machines. If we remove the moral human element that governs them in the name of expediency it could lead to untold consequences. Can you imagine a robot with a software glitch that turns on the soldiers it is supposed to protect because it mistook them for the enemy. Worst yet as these things grow and learn we will need to retain some form of control over them.
The problem is that armies of robots will be directed by an elite few and will have no ethical qualms about attacking anyone. Totalitarian regimes have always invented creative reasons that many, many ordinary free people who fail to bow in submission are “subversives” and must be eliminated. A reign of abject terror always results.
The robots will move rapidly into policing peacetime populations.
It is not so much the evil intentions of rulers that we must fear. It their impaired psychological and moral development and their unconsciousness of their own actions that will harm us. The adaptive psychopaths are the same ones with the control freak energy it takes to rise to and seize high level power. It takes a vigilance that humankind has never displayed to prevent this.
Robots have no autonomous conscience. They will make Hitler’s Third Reich look downright civilized.
Robots totally make sense for advanced ground and air scouts. I’m surprised there aren’t robots vehicles at the front of convoys to trigger off IEDs. How much control do you need - steering, pedals and gears? It might cost just a few thousand dollars to retrofit an old vehicle. The basic functions can be pre-programmed so the controlling personnel can focus on not accidentally driving over people.
It is easier to trust the robot than trust the politician who decides on where, when and against whom to deploy the robot. That is the real pandora’s box which is being opened.
Old techhnology is cheaper and can still do a better job.
You give me a 10,000 bi-planes that are fully armed along with a thousand good pilots and that squadron would put any drone to shame and at the same time do a thousand times the damage to terrorists.
Predators could be so effective protecting merchant ships from pirates off Somalia. Have some Predators always airborne in the area and when a ship’s captain radios that he’s under attack, the Predator flys to the ship, verifies the targets with its cameras, and with one hellfire missile, the pirates are vaporized. Seems so simple to me.
I have to disagree with the comments regarding the predictability and control-abilty of robot behavior. Predictability becomes harder as systems become more intelligent. For example: how can a programmer predict behavior when the program is allowed to change its own code to adapt to its experiences in the field (machine learning)?
I think that control-ability by an operator will become a bigger issue when threat identification by robots becomes better. Because receiving a go-ahead from a human could cause a significant delay in a life-or-death situation, I could see a compelling argument for eliminating this safeguard in some situations.
I am stunned at the retoric.
This is a great move. The saving of our boys lives is paramount and that is the goal. IT is being done well and all the other stuff is drivle and nonsense.