IEEE Technology and Society Magazine - June 2017 - 71

for protecting the public from vehicle-related harms.
While engineering standards are necessary, there are
many occasions where they are not sufficient. This is
saliently illustrated by the Pinto case where engineers
satisfied rear-end collision and other safety standards of
the time. Standards (whether from governmental entities
and/or professional engineering societies) and regulations (federal and state) often have difficulty keeping
pace with technological change, a challenge that is especially relevant to emerging technologies [46]. This seems
to be occurring in the case of self-driving cars where regulations and standards have indeed been slow to materialize as the technology rapidly develops [47].
Arguably, standards for self-driving cars would have
to be more rigorous than they are for traditional automobiles. Established "standards" for many of a car's
safety features (e.g., front/rear impact tolerances) can
be used to judge at least some of a designer's acts. But
self-driving cars bring into the picture added variables
for which standards must account; for example, a
designer would have to determine how to prevent users
from increasing the risk to which they expose themselves if the lack of control over the vehicle causes them
to override automated systems.
There will be major challenges even in the most optimistic scenario, in which the cars will have standard
interfaces that will encourage reliable interactions with
each other and with a central system for coordination.
In a less optimistic (but perhaps more realistic) scenario, automated cars will be developed by rival corporations that will be less interested in cooperation and
more interested in keeping their competitive advantage.
If designers are required to protect trade secrets and
market advantage while developing, testing, and maintaining their separate automated car, then achieving
system-wide reliability, and verifying that reliability, will
be all the more difficult.
And as was previously mentioned, auto manufacturers
are pursuing significantly different design pathways (e.g.,
Google vs. Mercedes). Thus, it will be difficult for regulators to develop a uniform approach to safety standards.
Among the crucial divides in the self-driving car industry
is whether a human being should remain in the driving
loop at least to some degree (Levels 2-4) or whether the
system should entirely take over the driving (Level 5).
There are conflicting opinions in the engineering community about whether humans should be "artificially
engaged" to keep their attention focused on a self-driving
car's functioning or whether the car should be fully autonomous [48]. If the former is pursued, then considerations
involving the interaction between the human operator
and the autonomous system such as Mean Time Between
Interventions (MTBI) and Mean Time to Intervene (MTTI)
are crucial for designers to address [49].
june 2017

∕

Another significant divide is whether the safety and
reliability of the car's functioning should be tied into an
ongoing communication stream between the car and
external systems (highway sensors, V2V, etc.) or whether the car should be "smart" enough so that it can operate independently from such input. This is sometimes
referred to as the distinction between "connected" versus "automated" autonomous vehicles [50]. Coordination among vehicles that adhere to these distinct
paradigms will be difficult.

An Appeal to Engineering Codes of Ethics?
One avenue for designers to obtain guidance on professional matters is through codes of ethics. The "paramountcy clause" from engineering codes (i.e., uphold
"the safety, health, and welfare of the public") is certainly well-intentioned and important, but it can be unclear
how to apply it to a particular case, especially when
there is a professional difference of opinion or there is
not much precedent on which to rely. Little specific
guidance is provided thus far by professional codes
regarding the design of self-driving cars or other "autonomous" technologies.
In general, beyond promulgating codes, professional
societies might be reluctant to actively promote "aspirational" ethical behavior in part because they may lack
consensus about which types of "good" behaviors
should be openly endorsed. Conflicts between engineering priorities and business priorities may also limit
the ability of professional societies to engage in ethics
promotion and support [51]. Yet aspirational behavior is
precisely what is needed in the case of self-driving cars
given how much of an effect the cars will have on the
lives and well-being of members of the public.

Moral Responsibility for
Computing Artifacts: The Rules
"The Rules," championed by Keith Miller in collaboration with other computer scientists, engineers, and ethicists were created with the intent of providing guidance
to the computing and engineering communities especially with respect to pervasive and autonomous technologies [9]. Unlike codes of ethics, The Rules do
provide specific guidance relevant to the design of selfdriving cars. The Rules are presented below with an
accompanying commentary for each one regarding selfdriving cars.
■ Rule 1 - "The people who design, develop, or
deploy a computing artifact are morally responsible
for that artifact, and for the foreseeable effects of
that artifact. This responsibility is shared with other
people who design, develop, deploy or knowingly
use the artifact as part of a sociotechnical system."
(emphasis added)

IEEE TEchnology and SocIETy MagazInE

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2017