Page 149 - Proceedings of the 2018 ITU Kaleidoscope

Machine learning for a 5G future

th
behavior etc. By the 17 century, ‗ethics‘ was accepted as Unlike Kantianism, the theory of Utilitarianism that
the science of morals, the rule of conduct, and the science originates from philosophies proposed by Jeremy Bentham
of human duty. Ethics is treated as moral principles that (1748-1832) and John Stuart Mill (1806-1873), examines
govern a person‘s or a group's behavior and actions. In right or wrong based on the consequences of an act or a
simple words, ethics refers to what is right and wrong and rule. The utilitarian viewpoint asserts that guiding principle
how to act on it. There is no single definition of ethics. of conduct should be greatest happiness of the greatest
Different definitions have their roots in the two basic number of people. In addition, Social Contract Theory ,
philosophies of realism and idealism, or objectivism and based on exposition and defense by Thomas Hobbes (1588-
interpretivism. The common thread in various theories and 1679), John Locke (1632-1704), Jean-Jacques Rousseau
definition is that ethics concentrates on human actions or (1712 -1778), John Rawls 1921- 2002), provides the
on the consequences of human actions. Thus, ethics may be justification for the establishment of moral rules to govern
defined as the systematic study of human actions from the relationships among citizens as well as the mechanism
point of view of their rightfulness or wrongfulness, as capable of governing these rules-governments [6].
means for the attainment of ultimate happiness for
individual and society [24]. 3. ETHICAL ISSUES IN MACHINE LEARNING

In this paper we examine the (i) the interplay of ML and ML is continuously unleashing its power in a wide range of
ethics; (ii) briefly discuss emerging ethical issues in use of applications. Traditional roles performed by humans are
ML that sometimes overlaps into AI; (iii) explore some now increasingly being taken over by machines and there
‗existent frameworks‘ for managing ethical issues arising in are also concerns being raised in the literature about human
ML; and (iv) suggest ‗new parameters to manage ethical workforce to be replaced by machines sooner than expected
issues‘ and provide ‗implementation strategies‘ as well as ‗ (for instance Rifkin, 1996) [39]. Google Assistant, inbuilt
way forward‘. with the feature of ―duplex‖ and IBM Debater are a case in
point - both depict the move to build technology adept in
2. SCHOOLS OF THOUGHT AND PHILOSOPHIES undertaking meaningful conversations or arguments with a
ON ETHICS human companion, thus partially fulfilling human need of a
companion [28] [8] but being ‗robots‘ they are capable of
The concept of ethics was originally proposed by the Greek ‗recording‘ and ‗analyzing‘ personal idiosyncrasies and
philosopher Aristotle for the discussion of philosophical behavioral preferences too.Similarly, ML based virtual
questions relating to daily life: the ‗ethike theoria’ that assistants ‗substituting‘ human beings in undertaking
deals with the study of, and gives criteria for the evaluation banking services have access to sensitive financial data. It
of human behavior. Since then, ethics has become one of is not an exaggerated presumption that such ‗machine
the major topics in Western philosophy [21]. Ethics always based assistance‘ would not only be susceptible to remote
change and they are handed down from one generation to hacking from outside but also can be equally prone to
the other. Therefore, what may have been correct or right or ‗instigation‘ of frauds from the machine ‗itself‘ [7] a la‘
good for one may not be the same for another? Life does ‗Winston‘, a conversational virtual assistant with AI
not always present us with the scenarios in black and white, abilities constructed to help Edmond Kirsch- the computer
rather we always encounter grey areas where every scientist, in the piece of fiction titled ‗Origin‘ (Dan Brown,
decision has an ethical implication and there is nothing 2017).
exactly right or completely wrong; for instance, the Indian
epic Mahabharata depicts complex characters who are The more we become reliant on machines, the more
fighting ethical dilemmas all the time. In general, ethics complex the relationship between humans and machines
involves the analysis of conduct that can cause benefit or gets. There are also questions being raised on whether we
harm to other people [4]. have gone too far when computer generated imagery (CGI)
can create realistic humans or when fake videos cannot be
In the western philosophy, there are many theories/schools distinguished from the real ones [27]? Big data empowers
of ethics viz., egoism, hedonism, naturalism and virtue ML algorithms to uncover more fine-grained patterns and
theory, existentialism, utilitarianism, contractualism and make more precise predictions than ever before [32].
religion [26]. However, most commonly discussed are few. Complicating the matters further is the vast quantity of
The first, the Virtue ethics inspired by Nicomachean Ethics easily available data. ML is continuously unleashing its
by the Greek philosopher Aristotle (384-322 B.C.E.), holds power in a wide range of applications. It has been pushed to
that virtues (such as generosity, charity and justice) are the forefront in recent years partly owing to the advent of
dispositions to act in ways that benefit the possessor and big data. ML algorithms have never been better promised
society [5]. Another related thread of rational, inspired by while challenged by big data. Big data enables ML
German philosopher Immanuel Kant (1724-1804), holds algorithms to uncover more fine-grained patterns and make
the concept of duty central to morality. Kantianism is more timely and accurate predictions than ever before [32].
concerned about not what we do, but what we ought to do. There is easily available data to the machine to learn from
What we ought to do reflects our dutifulness [4]. and over a period of time it will become difficult to

– 133 –