IEEE Technology and Society Magazine - June 2021 - 66

Feature
AI Ethics for Sustainable
Development Goals
Aníbal Monasterio Astobiza
Universidad de Granada, 18011 Granada, Spain
Mario Toboso
IFS-CSIC (Spanish National Research Council),
28037 Madrid, Spain
Manuel Aparicio
University of Murcia, 30003 Murcia, Spain
Daniel López
IFS-CSIC (Spanish National Research Council),
28037 Madrid, Spain
 We live in an era where problems are global
in scale (e.g., climate change) and solutions must
also be coordinated on a global scale in the international
context. The 2030 agenda for sustainable
development goals (SDGs) was ratified in 2015 as a
continuation of the millennium development goals
(MDGs). In this sense, the SDGs, as MDGs were in
their day, are a global mechanism that urges governments
to coordinate to address global problems.
At the core of the SDGs is " to achieve a better and
more sustainable future for all. " The SDGs consist in
a series of 17 goals with 169 targets which, for the
first time, identify the fight against poverty as a necessity
for sustainable development. The SDGs consider
the ecological, social, and economic dimensions as
interdependent for sustainable development. With
the progress and advances of artificial intelligence
(AI) technologies, many researchers are exploring
the possibility of their use to tackle societal problems.
This is what many people nowadays call " AI
for social good " (AI4SG). The concept behind AI4SG
is very simple: AI-powered systems and capabilities
applied to improve public welfare [1]. Although
there are different forms of classification of AI4SG
initiatives (in terms of data, modeling, or decision-making)
" projects addressing AI4SG vary significantly "
[2] and the AI behind these projects may
have been designed for the " good " but, in practice,
it could end up going " bad. " More importantly, not
everyone would agree on what is a good result. The
main motivation for any application of the AI4SG is
to solve social problems.
Digital Object Identifier 10.1109/MTS.2021.3056294
Date of current version: 3 June 2021.
66
0278-0097/21©2021 IEEE
However, the development of an algorithm in a laboratory
without the possibility of real-world application
is not AI4SG. The lack of AI research for proof of
concept or pilot testing with real-world application
and implementation often makes it very difficult to
evaluate many AI4SG projects. On the other hand,
at the background, there may be sincere motivation
and good intentions, but the application of AI may
create other social problems in addition to those
initially intended to be solved. For example, cities
around the world strive hard to reduce antisocial
and violent behavior, leading the city authorities to
rely on the use of AI systems that sketch the profile of
the population, that is, classify and predict people's
antisocial behavior. However, this has often resulted
in discrimination against certain groups. Algorithms
designed to predict antisocial behavior were fed
with not very robust data from statistics of illicit acts
leading to surveillance, and the simple mass surveillance
of these neighborhoods led to more violations
being reported for strictly statistical reasons which
in turn served to continue to feed the algorithm with
the consequent effect of self-fulfilling prophecy:
minority ethnic groups and stigmatized groups are
discriminated because of prejudice [3]. Preventing
this from happening is in the realm of ethics that
must be taken into account from the start to the end,
and engineers and computer scientists that want to
deploy AI technologies to solve complex problems,
first of all, bear ethics in mind while designing the
AI technologies. Obviously, not always everyone
knows what is involved in ethical design.
IEEE Technology and Society Magazine
IEEE Technology and Society Magazine - June 2021

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2021
