by Steven Metz
June 10, 2016
Navy Rear Adm. Mat Winter, left, and Navy Adm. Jonathan Greenert
with the Navy-sponsored Shipboard Autonomous Firefighting Robot,
Washington, Feb. 4, 2015
(Department of Defense photo).
"Fifteen years after a drone first fired
missiles in combat," journalist Josh Smith
recently wrote from Afghanistan,
"the U.S. military's drone program has expanded far beyond
specific strikes to become an everyday part of the war machine."
Important as this is, it is only
a first step in a much bigger process.
As a report co-authored in
January 2014 by Robert Work and Shawn Brimley
"a move to an entirely new war-fighting regime in which
unmanned and autonomous systems play central roles" has begun.
this ultimately will lead is unclear.
Work, who went to become the deputy secretary of defense in May
2014, and Brimley represent one school of thought about robotic war.
Drawing from a body of ideas about military revolutions from the
1990s, they contend that roboticization is inevitable, largely
because it will be driven by advances in the private sector.
the United States military must embrace and master it rather than
risk having enemies do so and gain an advantage.
On the other side of the issue are activists who want to stop the
development of military robots. For instance the United Nations
Human Rights Council
has called for a moratorium on lethal autonomous systems.
Nongovernmental organizations have created what they call the
Campaign to Stop Killer Robots, which is modeled on recent
efforts to ban land mines and cluster munitions.
Other groups and
organizations share this perspective.
Undoubtedly the political battle between advocates and opponents of
military robots will continue. However, regardless of the outcome of
that battle, developments in the next decade will already set the
trajectory for the future and have cascading effects.
points, autonomous systems will cross a metaphorical Rubicon from
which there is no turning back.
One such Rubicon is when some nation deploys a robot that can decide
to kill a human based on programmed instructions and an algorithm
rather than a direct instruction from an operator.
parlance, these would be robots without "a human in the loop."
In a sense, this would not be entirely new:
Booby traps and mines
have killed without a human pulling the trigger for millennia.
the idea that a machine would make something akin to a decision
rather than simply killing any human that comes close to it adds
greater ethical complexity than a booby trap or mine, where the
human who places it has already taken the ethical decision to kill.
"Creating autonomous military
that can act at least as
ethically as human soldiers
appears to be a sensible goal."
In Isaac Asimov's science fiction
collection "I, Robot," which was one of the earliest attempts to
grapple with the ethics of autonomous systems, an ironclad rule
programmed into all such machines was that,
"a robot may not injure a human
Clearly that is an unrealistic boundary, but as an
2008 report sponsored by the U.S. Navy argued,
"Creating autonomous military robots
that can act at least as ethically as human soldiers appears to
be a sensible goal."
Among the challenges
to meeting this goal that the report's authors identified,
"creating a robot that can properly
discriminate among targets is one of the most urgent."
In other words, the key is not the technology for
killing, but the programmed instructions and algorithms.
also makes control extraordinarily difficult, since programmed
instructions can be changed remotely and in the blink of an eye,
instantly transforming a benign robot into a killer.
A second Rubicon will be crossed when non-state entities field
Since most of the technology for military robots
will arise from the private sector, anyone with the money and
expertise to operate them will be able to do so.
Even if efforts to control the use of robots by state
militaries in the form of international treaties are successful,
there would be little to constrain non-state entities from using
Nations constrained by treaties could be at a disadvantage
when facing non-state enemies that are not.
A third Rubicon will be crossed when autonomous systems are no
longer restricted to being temporary mobile presences that enter a
conflict zone, linger for a time, then leave, but are an enduring
presence on the ground and in the water, as well as in the air, for
the duration of an operation.
Pushing this idea even further, some
experts believe that military robots will not be large, complex
autonomous systems, but swarms of small, simple machines networked
for a common purpose. Like an insect swarm, this type of robot could
function even if many of its constituent components were destroyed
or broke down.
Swarming autonomous networks
would represent one of the most profound changes in the history
of armed conflict.
In his seminal 2009 book "Wired
for War," Peter Singer wrote,
"Robots may not be poised to revolt,
but robotic technologies and the ethical questions they raise
are all too real."
This makes it vital to understand the
points of no return.
Even that is only a start:
Knowing that the
Rubicon has been crossed does not alone tell what will come next.
When Caesar and his legion crossed
the Rubicon River in 49 B.C.,
everyone knew that some sort of conflict was inevitable.
But no one
could predict Caesar's victory, much less his later assassination
and all that it brought. Although the parameters of choice had been
bounded, much remained to be determined.
Similarly, Rubicon crossings by military robots are inevitable, but
their long-term outcomes will remain unknown.
It is therefore vital
for the global strategic community, including governments and
militaries as well as scholars, policy experts, ethicists,
technologists, nongovernmental organizations and international
organizations to undertake a collaborative campaign of learning and
Political leaders must engage the public on this
issue without hysteria or hyperbole, identifying all the alternative
scenarios for who might use military robots, where they might use
them, and what they might use them for.
With such a roadmap, it
might be possible for political leaders and military officials to
push roboticization in a way that limits the dangers, rather than