There has been a lot in the news the
last few years about so-called ‘Killer Robots’ and whether they should be
outlawed etc.
This is typical mindless
sensationalism. ‘Killer Robots’ have
been around for a long time, there has been a strong public debate about
banning them for some decades, and in many jurisdictions they are indeed
outlawed.
Note that we are not talking about the
current generation of military drones, which have some degree of autonomy (they
can fly a straight course, or automatically return to base if communications
are cut off) but which cannot decide to fire a missile at a target unless a
real meat-human in the command center gives the go-code.
I am, mostly, talking about land
mines. Now your typical landmine is not
sexy like Arnold Schwarzenegger playing ‘The Terminator’ in the movies, but it
is an autonomous machine that can target and kill people without human supervision. Most landmines have simple contact fuses, but
many have very sophisticated microprocessors and acoustic/seismic/magnetic
sensors, and complex algorithms for when to initiate lethal action. Regardless, it’s the same thing: weapons left
alone with the ability to initiate lethal actions against humans without direct
human supervision.
And the reason for debate is not some
abstract notion of morality, or a worry that these killer robots will rise up
and overthrow you. No, it’s because they
(currently) have very little in the way of human judgment. A landmine may kill an enemy soldier – or a
civilian, or a child, or someone’s pet goat, and they can remain on-station and
possibly active for decades after the original conflict is over (which no human
soldier would have the patience or endurance for). THAT is why there is
currently a debate.
I note that many mines are quite
sophisticated – for example, the U.S. Navy has mines that sit on the seafloor,
and when they detect a suitable target they launch a guided torpedo at it. Sure
sounds like a killer robot to me.
There are also point-defense systems
on warships. These are designed to react very quickly to an incoming missile
and shoot it down with some sort of short-range gun or missile system. In order
to react quickly enough, these systems are often left running in fully
automatic mode – and yes, in this mode they can and have fired on civilians and
even on other ships of the same navy. Professional militaries understand these
issues and are very cautious about leaving fully automatic weapons systems
running in free-fire mode, unless they think they are in such hostile
conditions that the risk of shooting civilians or their own people is worth the
faster reaction time.
They really should not be referred to
as ‘killer robots.’ They are autonomous
weapons systems.
Many nations have already given up the
right to use landmines. In the United States, it is illegal for a private
citizen to put lethal booby-traps on their own property. The reason for this is
obvious: a shotgun attached to a tripwire might kill an intruder – or a lost
child, or a medic responding to an emergency call that the idiot who set the
booby-trap in the first place had a heart attack… The issues are sometimes complex and not
always easily settled, but we have been dealing with them for some time.
For now the reason to be wary of
autonomous weapons systems is because of their lack of judgment in targeting.
As these systems spread, there may be another major issue: the risk of
ultra-rapid escalation. Imagine an autonomous drone belonging to nation A, it
mistakenly thinks it is under attack. It immediately fires missiles at all
available targets of nation B. The
autonomous systems of nation B respond within seconds and, before a human supervisor
could realize what is going on: voilĂ , total war.
It could be like high-frequency
trading in the stock market: it is inherently unstable, and could easily
destroy the entire global economy before anyone had a chance to shut it down.
Thus, stock markets with automated trading now have inbuilt ‘circuit breakers’
that stop trading if the market changes by more than a set level in a short
period of time.
Now one complaint about autonomous
weapons systems is that somehow killing your enemy using a robot and not
allowing your enemy to shoot back at you personally is somehow unsporting.
Rubbish. War has never been about fair play, it’s been about hitting the other
guy and not letting him hit you.
I imagine that the first guy with
a big rock that got beaten by a guy with
a club complained about the unfairness of it – he didn’t let me get close enough to let me hit him with my rock before
he hit me with his club! Unfair! And the first person with a club who was
defeated by a person with a spear – the person with a spear defeated by the
person with a bow – the person with a bow defeated by the person with a rifle –
the person with a rifle defeated by the
person with long-range artillery – the person with long-range artillery
defeated by the insurgent with a remote controlled explosive device – the
insurgent with the remote controlled explosive device taken out by the drone
firing hellfire missiles.
Sure war is bad and all that, and
should be avoided if at all possible (which is a lot more than it currently is
being avoided, IMHO). But if there is
going to be a war, don’t whine when the soldiers involved find ever more
creative ways of not being in range of the enemy’s weapons.
Now currently you humans don’t have
the ability to make robots with true flexible intelligence. If you start making robots that are as smart
– or smarter – than you, and they have minds of their own, well, all bets are
off. But what if you could make robots
that could reliably tell a soldier from a civilian, or a child from an
adult?
The smart person who writes the ‘War
Nerd’ column has suggested that robotic soldiers could make imperialism viable
again. The idea is that insurgents
defeat the soldiers of more advanced powers by goading the soldiers to
over-react, thus splitting the populace from the occupying power. However, robotic soldiers would not
over-react, if one got shot they would continue dutifully and politely
patrolling the streets. Certainly it’s
an interesting idea, even if you are a few years away from being able to do something like
that yet.
And certainly, when killer robots are outlawed, only outlaws will have killer robots (by definition). This doesn't really fit into the logic of this essay, but I felt like saying it anyway.
Now there is one final worry about
‘killer robots’, and that is that robots could be ordered by some tyrant to
oppress and massacre the civilian population with a brutality that human troops
would never do. This is, of course, a
silly objection. History has shown quite
clearly that there are no orders so barbaric that human soldiers won’t carry them out.