IEEE Technology and Society Magazine - March 2021 - 60

AI for a Fair, Just, and Equitable World

other domains using AI systems, human resources,
the management or executive boards, and the end
users. As a society, we ought to demand the inclusion of values for a transparent and responsible
human-centered AI that:
*	 puts upfront the respect for fundamental human
rights over performance or economic benefit
and ethical requirements such as transparency
and accountability are key to promote responsible practices of AI and enhance trust in society;
*	 provides means to empower people interacting
with AI by educating users and today, STEM strategies are not enough; it is also necessary to build
upon AI and data literacy to up-skill or re-skill
citizens, to allow them participate in the social
and economic changes to come [49], as well as
to understand the capabilities and limitations of
new technologies;
*	 are bound to regulation without killing innovation and governments need to create networks
of collaboration among research institutions and
industry to retain talent and innovation, while
building owned technology that strengthens the
socioeconomic and cultural ecosystem.
The main question remains, how can AI systems
be designed to align with ethical, legal, and societal principles? AI-based systems are only as good
as the data and rules designers feed into them. If
the data is biased or incomplete, then the decisions taken by the AI programs are also biased or
are least inaccurate. Bad data results in codifying
our implicit racial or gender biases into AI-based
system programs. It is, therefore, an imperative that
data, rules, and training sets used for building AI systems ought to be unbiased and unconscious preferences of the AI designers should not seep into
training data [1]. It is clear that having more computation does not make us more intelligent, only more
computationally powerful.

Ethics and AI in public services
AI will have a key role in the digital transformation of public services and administrations. In
the " AI digital transformation in public services "
section, we have highlighted some examples of AI
applications in the main domains of public services
that are already implemented and used in practice.
However, the more these technologies are being
used by the public administration in automated

60

decision-making or law enforcement, the more ethical issues are raised, as the explanation for how
and why an AI application reached its conclusions
is still unclear. The opacity of AI systems in any area
of public administration challenges the traditional
responsibilities of administrators regardless of substantive policy area or whether decisions directly
involve citizens. The core issue that is raised involves
the accountability of the AI system and the responsibility of the administrator who manages the program
in which AI is used [50].
Governmental data controllers must notify citizens which personal and sensitive data is being
gathered, stored, and processed, how their data
will be used, including the existence of automated
decision-making, and, at least in those cases, meaningful information about the logic involved [the
algorithm(s)], as well as the significance and the
envisaged consequences of such processing for the
data subject. In other words, public administrations
have the obligation to protect citizens from any possible harm caused by the AI system, while leveraging
the benefits of computational and economic benefits of such solution, or as explained in [51], to govern algorithms while governing by algorithms.
One example of the controversial use of AI solutions in the public sector is the rollout of surveillance
systems using facial recognition technology. With
the final purpose of increasing public safety, these
systems are collecting a massive volume of personal
data that are affecting citizens' privacy. Moreover,
there have been cases in which the algorithm has
showed to be biased, ending up in the discrimination of vulnerable groups of the society, generally
people of color or other social and demographic
minorities [52]. This same tradeoff between surveillance and privacy has been a constant topic of discussion during the COVID-19 pandemic, where the
rush to find solutions to help health care agencies
to mitigate the spread of the virus led to the appearance of digital contact tracing technologies that use
sensitive data such as the localization or the Bluetooth signals of mobile devices to identify potential
infected citizens along with their contacts and to
apply preventive measures. Again, in this case, the
proposed solution can leave out different social
groups that will not benefit from the services provided by such technology as they do not have access
to it. In addition, it might be the case that citizens
are more willing to provide personal data in these

IEEE Technology and Society Magazine
IEEE Technology and Society Magazine - March 2021

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