come more thoughtful and better informed about their work and its long-term effects.
As in other areas of thought, this
viewpoint diversity is a strength when it
can be harnessed toward a productive
exchange of ideas and perspectives. An
example of such an exchange is the ongoing debate within the artificial intelligence research community about the
appropriate value systems on which to
build artificial intelligence systems. The
goal of teaching ethics is to foster the debates and equip practitioners to participate productively. It does so, not by imposing a value system on students, but
by informing them about the range of
ethical descriptive and evaluative tools
available to them.
At the same time, educators should
make students and professionals aware
of the social ramifications of their work,
that research, development, and implementation can be carried out in a variety
of ways and for a variety of ends. Computer science educators should dedicate
significant time to ethics education,
helping enable students to make informed, thoughtful, ethical choices
about technology and its applications
and implications.
What is ethics? Ethics can be understood as the task of answering “What
should I do?” which is never a simple
matter. Ethics includes both thought
and practice, an organized and intentional reflection on morality and the effort to live in ways that are good, just,
and/or right. Although many people use
the words morality and ethics interchangeably, many ethicists understand
them to be different. One common way
of drawing the distinction is to define
“morality” as a set of values or a worldview and “ethics” as the practice of reflecting on those values and their foundations and applications. 4, 22
There are many different, often con-
flicting, ways to understand how to be
moral. The clashes are sometimes be-
tween people who share the same funda-
mental premises and method of inquiry
into how to be moral but disagree about
conclusions. Other times, the clashes
are between people whose basic ideas of
how to answer the question of how to be
moral conflict with one another. Most
approaches to morality can be under-
stood in terms of the three major tradi-
tions of ethical thought—deontological
ethics, virtue ethics, and utilitarian-
ism—with each growing out of different
core questions and ways of seeing the
world.
Ethics is typically understood to be
normative; that is, it is aimed at establishing norms of thought, values, or conduct. This assumption is especially prevalent in professional ethics courses that
are typically used as a means to steer students’ future behavior toward a set of
professionally agreed-upon values (such
as professionalism and honesty). 26 But
ethics is also a tool for description, furnishing decision makers with a critical
framework that enables them to understand what is happening in a given situation and what is at stake in any action
they might take. The boundary between
normative and descriptive functions is
sometimes fuzzy; for example, it is often
the case that different details of a situation will appear salient depending on
which ethical approach one adopts. This
malleability of relevant details can make
ethics itself seem murky or imprecise.
However, teaching students to appreciate this difference, understand the
modes of reasoning that they or others
might employ in making an ethical decision, and move between these reasoning
structures themselves is the goal of a
good ethics course.
Educating students in the descriptive functions of ethics is as important
as communicating to them the professional norms of computer science.
Computer science is a field in which
everyday practice and problem solving
takes place in a context that could barely be imagined the decade before. Educators cannot predict the ethical quandaries their students will face. With an
education in ethical description, the
students will be better able to engage
in subtle and substantive ethical reasoning when new and challenging
problems confront them.
Practical challenges of teaching
ethics. Ethics education is a notable
challenge for two reasons. First, in the
absence of any ideal universal ethics
program, students must be taught how
to approach problems as distinct from
being led to particular pre-ordained conclusions that might narrow their vision
and exclude important elements of a
given problem. Second is how to achieve
this goal while overcoming the biases
students often bring to the classroom.
how to choose. Callahan10 also endorses
the idea of helping “ ... students develop
a means and a process for achieving
their own moral judgments” when confronted with challenging situations.
It is essential that open ethical debates between well-informed practitioners take place. Computer science does
not take place in a vacuum; to an ever-increasing degree, the IT systems and
platforms, from search engines to
smartphones, that are built by computer scientists and engineers are creating
and redefining the social, political, and
individual contexts in which human beings understand themselves. 21 Whatever principles and norms are adopted by
computer scientists, and reinforced
through the design and deployment of
their systems, will have profound ethical and societal implications. Teachers
and leaders in the field have a responsibility to drive the discussion about the
effects of their own work and the work
of their students. Indeed, Boyer6 argued
that academics have a responsibility to
engage students and the public with
their research.
We have started to see this engage-
ment through a number of initiatives in
the computer science community, in-
cluding the International Joint Confer-
ence on Artificial Intelligence 2015 letter
on autonomous weapons researchc and
the 2017 follow-on letter signed by CEOs
of tech companies around the world;d
ACM statement on algorithmic
accountability;e development of the
IEEE standard for algorithmic bias
considerations;f and new conferences
and research groups focused on fair-
ness, accountability, and transparency,g
as well as conferences focusing on the
effect of artificial intelligence on soci-
ety.h These debates are important for
shaping the direction of the field, even
though they rarely result in consensus.
The utility of the debates is not that they
result in standardized practices but
rather that individual practitioners be-
c http://futureoflife.org/AI/open_letter_autono-
mous_weapons
d https://futureoflife.org/autonomous-weap-
ons-open-letter-2017
e https://www.acm.org/binaries/content/
assets/public-policy/2017_usacm_statement_
algorithms.pdf
f http://sites.ieee.org/sagroups-7003/
g http://www.fatml.org/
h http://www.aies-conference.com/