IEEE Technology and Society Magazine - Spring 2013 - 62

much more difficult to anticipate
the resulting consequences.
A significant problem is that
robots could fail in a variety of
ways. roger Clarke pointed out
years ago that "...risks include
trapdoors, Trojan horses, and similar devices in the robot's programming" [18]. He also claimed that,
"Complex systems are prone to
component failures and malfunctions, and to intermodule inconsistencies and misunderstandings"
[19]. Surely as robots become more
sophisticated, the complexity of
their underlying software and control systems should give us pause.
For example, an alleged virtue of
military robots is that if an error
is discovered, it can potentially
be systematically eradicated from
a whole line of robots at the same
time. On the other hand, it is possible that the vulnerability could be
exploited in each instance of a particular robot of that type (at least
until the issue is detected and corrected). The breaching of a robot's
security may be just the beginning
of a problem. After gaining control
of a robot, hackers could then use
that robot (or group of robots) for
malicious ends.
In general, the knowledge and
skill used to discover and eradicate vulnerabilities in an electronic
technology could also be applied to
breach its security. This is a widely
known issue in computing communities, illustrated by the ongoing
battles of individuals on different sides of the "white hat," "gray
hat," and "black hat" divides [20].
The term "dual use" is instructive
here; it roughly refers to the notion
that knowledge and technology
can have both beneficial and malicious applications.2 A slightly different interpretation of the term
is described by Nagenborg et al.,
who suggest that dual use "...is the
possibility to use civil technologies for the purpose of war" [21].
The InterAcademy panel (IAp),
a collection of scientific societies
from several countries, released a
62

position statement about the obligation that scientists have to consider
"dual use" applications, which
indicates the importance and relevance of this general issue [22].
The "dual use" issue should certainly resonate with academics and
practitioners now that the National
Institutes of Health (NIH) and
other agencies had to determine
whether to allow the publication of
papers relating to how the bird flu
virus could be altered to become
more infectious [23].
The possibilities for dual use
are rather extensive. For example,
understanding what would seem to
be the relatively innocuous behavior of fruit flies could potentially
lead to military applications [24].
Also consider the 2009 attack
launched against the Israeli navy;
purportedly, the incident involved
a remote control and a boat loaded
with explosives [25]. Further,
mobile phones have been used to
ignite improvised explosive devices
(IEDs), illustrating that a common
technology can be "re-tasked" for
destructive aims.
Drawing attention more specifically to the intersection of robots
and dual use, it will be exceedingly difficult to anticipate all of
the ways in which robotic systems
could be manipulated or reprogrammed to achieve malevolent
ends. Sharkey, Goodman, and ross
describe several plausible scenarios
where robots could be used to commit a crime, such as by unlawfully
entering a building through a window and stealing information [26].

2
The National Science Advisory Board for Biosecurity (NSABB) was created to assist the federal
government with developing policies in the realm
of dual use. The NSABB's working definition
of "dual use research of concern" is as follows:
"...based on current understanding, can be reasonably anticipated to provide knowledge, products, or technologies that could be directly misapplied by others to pose a threat to public health,
agriculture, plants, animals, the environment,
or materiel." Available at http://oba.od.nih.gov/
biosecurity/nsabb_faq.html#NSABB_FAQ001,
accessed Oct. 16, 2012.

The versatility of computing in
general and robots in particular is
a major strength of these technologies, but that versatility can generate unintended and unanticipated
consequences. Vacuum cleaning
roombas have been reprogrammed
to play a version of pac-Man [27].
It is not hard to imagine ways in
which this rather inexpensive platform could also be repurposed for
harmful acts. The company that
created the roomba, irobot, also
designs robots for the military, and
some of those robots carry firearms. The creation of the aforementioned version of "roomba
pac-Man" was motivated to test
aerial drone software [27].

Robots and the Law
The possibilities for robot misbehavior have naturally generated
interest among lawmakers, lawyers, and other members of the
legal community. ryan Calo points
out that a crucial legal difference
between computers (like a pC) and
robots (like a roomba) in that a
robot could inflict direct physical
harm, which is not at issue with a
stationary computer [28]. Calo contends that uncertainties about legal
liability for robot software could
reduce innovation and investment,
and he calls for laws that "shield
manufacturers and distributors of
open robotic platforms from suit
for what consumers do with their
personal robots, just as [Congress]
immunizes gun manufacturers
from suit for what some people do
with guns" [28].
The article by Calo and one by
Weng and Zhao [29] emphasize
that law (in many countries) about
these issues is at an early stage of
development. Both reflect the need
to balance the economic harm
that could result from increased
liability for developers, and the
physical harm that could result
from decreased liability for developers. Both articles maintain that
it is important for governments and
the public to engage in informed

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

SprING 2013

http://oba.od.nih.gov/

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - Spring 2013