DIGITAL PHRENOLOGY: ALGORITHMS AND ETHICS FROM A SYSTEMS PERSPECTIVE

Abstract

Decisions made by machine learning algorithms have societal impacts, both good and bad. This study acknowledges the great good that machine learning algorithms contribute to society while also pointing out that there are negative im-pacts such as bias, discrimination and misrepresentation of data, etc. Algorithms often use either irrelevant or incomplete data to make life-changing determinations about certain groups of society. The algorithmic use of data that does not repre-sent the true or actual ability of individuals or groups of people leads to digital phrenology. When algorithms are designed and tested outside the real world, it may not be obvious that digital phrenology may be caused. The ethical conse-quences of this problem can be mitigated if algorithms are not viewed as separate from the world it is eventually deployed into. Therefore, a more holistic systems view of algorithms, including their ethical consequences, should be taken. The purpose of this paper is to show that digital phrenology exists and that, once de-ployed, algorithms become part of a societal system where ethical values cannot be ignored. Various efforts and approaches to mitigate the ethical concerns related to unethical algorithmic behaviour have either done well in raising awareness or produced promising results. However, very little attention has been given to the explicit use of ethical theories as a possible approach to alleviating the problem. Therefore, two major ethical theories are proposed to assess the ethicality of hypothetical machine learning scenarios. A critical analysis of related literature and case studies is central in this study's inquiry method. Results show the existence of digital phrenology. It also reveals how viewing algorithms as a single compo-nent of a larger societal system can give a better perspective on the possible ethi-cal consequences that arise from their use. This research contends that the use of ethical theories can contribute to addressing digital phrenology by viewing algo-rithms as one of many components in a larger system.