IEEE Technology and Society Magazine - March 2018 - 60

Opponents of the ban [2], [5], [18] argue that we can't
predict what future autonomous systems may be capable
of. They might even outperform humans in many tasks
related to discrimination. Nevertheless, with present-day
technology, even when considering state-of-the-art systems, it seems Sharkey's third objection - machines
lacking the contextual awareness necessary for discrimination decisions - is a difficult one to solve.
Many weapons were banned because they failed the
test of the principle of discrimination at the time they
were banned. Arguing that autonomous weapons
should not be banned today because they might get
better in the future requires a leap of faith and it is akin
to arguing that the ban on antipersonnel landmines
should be lifted because in the future they might be
equipped with specific sensors that will discriminate if
the person stepping on it is a lawful target or not. The
ban on antipersonnel landmines considers them indiscriminate weapons in its current technology, it doesn't
take into account hypothetical improvements that might
or might not arrive.
Until there is serious evidence or research that suggests that an autonomous system has the capacities to
make discriminating decisions in complex environments such as an (urban) battlefield, AWSs must be
assumed to fail to comply with the criterion of discrimination, a crucial principle in both just war theory and
IHL, and should therefore be banned.
Schmitt and Thurnher [2] state that "more may not
be asked of autonomous weapon systems than of
human-operated systems," referring to the fact that
humans can also fail to pass the test, and that "enemies
have been feigning civilian or other protected status [...]
for centuries." Although this is a strong point, it remains
unconvincing as an argument against a ban as it only
shows that humans do not have a 100% success rate at
passing the test. The question remains if autonomous
systems equipped with current technology are at least
as good as humans in discriminating civilians from
combatants. This issue could be solved empirically, but
meanwhile we could safely guess that humans are
much better than today's systems, which as we have
seen, have trouble determining if a sausage pizza is vegetarian or not.
There are, of course, many other consequentialist
and deontological arguments against and in favor of
AWSs, see [2], [3], [11], [19] for more of these.

A Note on the Accountability of Killer Robots
Considering that a ban - for political, military, commercial, or other reasons - might be difficult to achieve
[4], it is perhaps inevitable that deploying AWSs eq uipped with today's technology could result in many
unjust deaths.

Defenders of autonomous weapons argue that
humans too cause unjust deaths. It would be counterfactual to deny that humans do indeed commit atrocities
during war, either by mistake or with intent. So what is
different with AWSs? Atrocities that are committed with
intent are called war crimes and the responsible individuals are meant to be held accountable under IHL.
Accountability, i.e., allocating individual responsibility for
war crimes, serves as deterrent of future harm, as an
exercise in retribution for victims, and it is a necessary
condition for fighting a just war [3], [19].
However, many questions arise in regard to AWSs
and accountability: what happens when an autonomous
weapon commits a war crime? Who should be held
responsible for an unjust action carried out by an AWS?
These and other questions were explored elsewhere [4],
[9], [19] in more detail, but I wish to join the conversation in this last section.
The whole notion of accountability becomes fuzzy
with an AWS, as probably neither the system's designers or developers nor the maintainers or the military
who deployed the system wilfully acted with the intention of causing that crime and are, even, bound to be
unaware of the concrete particular situation where the
unjust action took place. Sparrow [19] has argued that
a programmer or commanding officer are not satisfactory loci of responsibility because it would be unfair to
hold someone "entirely responsible for actions over
which they had no control." We should keep in mind
that we are not discussing weapons, but weapon systems, the focus does not lie on the weapon qua instrument, but on "the process by which the use of force is
initiated" [11].
Without intent there is no crime, so who should be
held responsible? The question is very difficult to
answer, but the consequences of leaving it unanswered
can be disastrous. A blurring of accountability "would
fail to deter violations of international humanitarian law
and to provide victims meaningful retributive justice" [9].
Although jus in bello requires a clear accountability
chain [3], Anderson and Waxman [5] do not consider the
accountability gap to be sufficient reason to block the
development of AWSs, as their use could end up saving
lives. Sharkey [1], on the contrary, argues that without
clear accountability to enforce compliance many more
civilian lives could be lost. Others argue that no such
gap would ever exist as there will always be a human
involved, at least at the point of deployment, who could
be held responsible [14].
I shall not elucidate this conundrum here. It is clear,
however, that the issue of the accountability of autonomous weapons, in situations when no person involved is
acting with the intent of committing a war crime, should
be formally addressed assuming that AWSs might be

IEEE Technology and Society Magazine

∕

MARCH 2018

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2018