IEEE Technology and Society Magazine - March 2021 - 58

AI for a Fair, Just, and Equitable World

before. Conversely, the use of AI in the security and
military areas is the one that generates more controversy and mistrust in society, not without reason. The
use of autonomous weapons capable of searching
and eliminating a target without human intervention
generates a concerned debate [30], [31], not only
for the profound ethical implications of leaving the
decision of taking a human life to a robot but also
because of the possibility of errors. Who is responsible for a death generated by AI? The developer of the
technology? The government that puts it into operation? Machines should not be allowed to decide to
kill a human being or decide who has to die.
Justice: The application of AI methods providing support to judicial decisions can help improve
the consistency of sentences and the predictability
of law application. These tools can be designed,
for example, to provide general legal advice to the
public, provide orientation based on jurisprudence
to law professionals, or assist in court legal decision-making [32]. In the United States, AI is already
being used in criminal justice across the country, for
instance, to support making bail and parole determinations with products like Correctional Offender
Management Profiling for Alternative Sanctions
(COMPAS) [33].
The research carried out by the European Commission for the Efficiency of Justice (CEPEJ) [32] has
mainly identified six categories in which AI is used
in the judicial environment: 1) advanced case-law
search engines, 2) online dispute resolution [34];
3) assistance in drafting deeds; 4) analysis (predictive and scales); 5) categorization of contracts and
detection of divergent or incompatible clauses; and
6) chatbots to inform or support litigants in legal
proceedings.
The implementation of predictive justice and AI
in the judicial system has many and multifaceted
implications, especially those related to ethics, fairness, and potential biases, which will require a prudent and balanced approach by decision-makers.
For example, the study carried out by Propublica on
the COMPAS tool mentioned above found that black
individuals were far more likely than white individuals to be incorrectly judged to be at a higher risk
of criminal recidivism. In contrast, white individuals
were more likely than black individuals to be incorrectly flagged as low risk of recidivism. When fundamental rights like freedom of individuals are at stake,
sensible decisions that are based on AI decision

support systems require guarantees that those systems are ethical, fair, unbiased, and sound.
Social services: The provision of social services
can also benefit from the AI revolution by enhancing
the ability to detect patterns in the population and
more easily identify citizens at risk who require public intervention and assistance. The potential offered
by ML, NLP, image analysis, and decision support
systems provides a range of tools to deal with the
massive processing of large amounts of structured
and unstructured data-in written reports, social
networks, photographs, interagency databases-that
are currently difficult to address manually, allowing
case managers to spend more time interacting with
people. Social services can benefit significantly
from AI-based tools to significantly improve the lives
of vulnerable populations. For instance, there are
already studies that provide significant advances in
predicting suicides, allowing health and social services to anticipate a suicide attempt and provide palliative interventions in time [36]. Other studies have
used ML techniques to predict episodes of domestic
violence, enabling the state to provide the necessary
means in advance [37]. Social exclusion detection
has also been addressed with AI tools [38], trying to
provide novel methods to social workers to anticipate to chronification of social exclusion of vulnerable citizens.
The application of AI in social services is a paradigm of AI for good. However, even when the intention to do good is necessary, it is insufficient, we
must take into account that in some situations the
misuse of these tools, even unintentional, can create or amplify other risks or harms. For example, the
use of biased data or methodologies poses a considerable risk that can generate new discrimination or
unfairness situations. Trustworthiness is essential in
technologies, in general, but is especially relevant in
technology for social good.
Environment-energy and water management: AI
is becoming more and more relevant in the environmental sectors, which, in some cases, are still participated by public administrations in many different
ways. The possibility of integrating AI techniques into
the management of energy networks offers enormous
potential for improving their efficiency and security
impacting in citizens' lives. With the present progressive digitization of the distribution networks (power,
gas, and water) and the integration of Internet of
Things (IoT) technologies, it is much easier to capture

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