IEEE Technology and Society Magazine - June 2016 - 43

would be saved due to the switch actuation even if the
one were not on the diversion track [21]. Therefore,
there has been no violation of a human right to be protected from involuntary use of their body or work for
the greater good of others. In other words, the one person being sacrificed is dying as a matter of collateral
damage. He is not being "used" to save the five. Of
course, the importance of this distinction is likely to be
lost on the one being killed, since he could justifiably
feel that he has a right not to be sacrificed without his
consent. In fact, according to libertarian philosophers
Vallentyne and van der Vossen, a core tenet of libertarianism holds that persons have a right of full self-ownership. That right of self-ownership is violated when an
individual is subjected to non-consensual and unprovoked killing, maiming, enslavement, or forcible
manipulation [22]. Additionally, the fact that such a
nuanced distinction, between collateral human sacrifice and unfair use, might be offered - in an attempt
to blunt the troubling ethics of sacrificing others -
underscores the need to carefully vet technologies that
would electronically adjudicate these sorts of unsettled
ethical controversies.
We can imagine trolley-like scenarios for automobiles.
Two young children dash into a street before an oncoming vehicle, one chasing the other. The driver has no time
to brake yet can swerve just enough to miss them - but
only by hitting a single bystander child on the road shoulder. It is one thing for a human driver to be faced with
such an awful choice and live with the consequences. It
is something entirely different for choices like this to be
preprogrammed into a machine, thereby encoding mandates to ethical questions that are far from settled in our
civil polity. In the first case, someone onsite makes a
heart wrenching ethical decision and takes responsibility.
In the second, someone offsite makes an advance calculation without witnessing the effects, and attorneys fight
about responsibility in the aftermath.
During the Automotive Tech.AD 2015 conference I
posed the above "two-vs-one" child scenario during an
"ice breaker" round table discussion, and was not surprised to hear the utilitarian reply. All who responded felt
that running over one child was preferable to running
over two, even though it would involve a choice to purposefully steer into the one. These were senior managers and executives from companies directly involved in
the development of autonomous vehicle technologies,
and their replies reflected a social utility that most -
though not all - of us intuitively accept. We want to minimize the sum total of human suffering, though we are
often unsettled as to what constraints should be applied
to the means by which we achieve that goal.
As a follow up to their utilitarian reply, I offered a description of another - this time non-automotive - ethical choice
june 2016

∕

that is championed in a journal paper coauthored by Katarzyna de Lazari-Radek and the famous utilitarian philosopher Peter Singer. In that paper the authors present a
hypothetical scenario about a brain surgeon operating
on a patient [23]. The patient is an ideal organ donor.
There are four other dying patients who could all be
saved by the organs of the surgery patient - each via
receipt of a different organ. Singer and de Lazari-Radek
argue that, according to utilitarian consequentialism, the
brain surgeon would be right to kill the patient being
operated on and use her organs to save the four in waiting - provided that the surgeon can do so without
being discovered and does not publically condone the
act or publically promote acts of this nature [23, p. 40].
This "esoteric morality" is a coherent - albeit extreme
- extension of utilitarian consequentialism, provided
that absolute secrecy of the act can be maintained.
Singer and de Lazari-Radek note that secrecy is needed
(in part) to avoid negative consequences regarding loss
of faith in the medical profession if doctors are known
to kill some patients to save others. They acknowledge
that keeping these sorts of secrets is difficult, and also
that doctors who carry out such practices are likely to
lead themselves into more risky moral judgment calls.
Therefore, they admit we should advise doctors against
such practices even though the execution by surgery
would be the morally correct thing to do - so they
claim [25]. The folks that I shared a round table with at
the conference were quick to condemn the notion of a
doctor sacrificing one patient to save many. Their selfconflicting reactions to the autonomous vehicle steering into the child versus the murderous doctor
epitomize the ambivalence that we as individuals and
as a civil polity experience when confronted with the
insoluble tension between serving the greater good and
respecting the rights of individuals.
A notion of duty toward vehicle occupants offers yet
another ethical challenge to those who are developing
autonomous vehicles. Would it be acceptable for an
autonomous car to swerve into a lamppost thereby sacrificing you, its sole occupant, in a utilitarian effort to
avoid the two children in the road? When you engage a
self-driving vehicle to transport you, is there a duty -
born by the vehicle's designers - to protect you above
all others in all circumstances? Suppose that you and
your own child are on board. The utilitarian calculation
must consider the greater probability that you and/or
your child might survive the lamppost as compared to
the odds against the two children on the street. Is it ethical for engineers programing vehicles to take those
sorts of chances with the lives of occupants if the greater good is served?
Examples of children at dire risk in the street are, of
course, rare in terms of our daily driving experience.

IEEE Technology and Society Magazine

http://www.Tech.AD

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2016