Bias and Discrimination in Artificial Intelligence: Emergence and Impact in E-Business

Introduction

In the course of the advancing digital transformation, artificial intelligence (AI) has recently become increasingly important for society, economy and government. The application of AI particularly opens up a variety of new opportunities for organizations in e-business, promising considerable gains in effectiveness and efficiency in the execution of business activities and processes. In e-business, AI is, for instance, already used to determine which and how online ads are presented to a person (Hillard, Schroedl, & Manavoglu, 2010), the order in which consumers are presented the news (Nalisnick, Mitra, Craswell, & Caruana, 2016), whether a person is hired for a job (Strohmeier & Piazza, 2015) or whether a person is eligible to get a loan (Khandani, Kim, & Lo, 2010). Accordingly, AI plays a significant role in e-business and our lives, increasingly impacting our opinions and behavior in everyday life (Leavy, 2018). To put it in a nutshell, AI is regarded as a key technology and economic growth engine that is essential for the future sustainable competitiveness of e-businesses.

At the same time, however, there are major risks and challenges associated with the application of AI in the context of e-business, which may be technological, legal, social and ethical in nature (Wirtz, Weyerer, & Geyer, 2019). An intensively discussed aspect refers to the ethics of AI and the question of whether the development and application of AI and its consequences are ethically and morally justifiable or how ethical basic values can be integrated into AI applications (Anderson & Anderson, 2011; Lin, Abney, & Bekey, 2012; Wirtz et al., 2019). A significant ethical risk refers to the issue of AI-related bias and discrimination, which is increasingly attracting public attention and has become subject to a great social debate. AI-related bias and discrimination refers to the fact that AI can adopt human prejudices or discriminatory values and behaviors and even reproduce them in an amplifying way (Basu, 2018; Thierer, O’Sullivan Castillo, & Russell, 2017).

In general, this means that AI applications may potentially not only violate national anti-discrimination laws implemented by a great number of democratic governments, such as the U.S. (e.g. The Age Discrimination in Employment Act of 1967.), UK (e.g. Equality Act 2010.) or Germany (e.g. General Act on Equal Treatment.), but also the prohibition of discrimination enshrined in Article 2 of the Universal Declaration of Human Rights.

The issue of AI-related bias and discrimination and its potentially adverse effects on people and society is particularly relevant and far-reaching in the e-business realm. Here, AI’s scope of application is extremely high and constantly advancing, yielding new, more and more pervasive AI applications, which affect a particularly large number of individuals – not to say our whole society.

Against this background, awareness of the topic among consumers has increased, putting the beneficence of AI into question. As a consequence, the acceptance and usage of AI solutions in e-business may be at stake in the long term, if the concerns and uncertainties among consumers further disseminate and solidify, thus threatening the great potential of AI for organizations in e-business. As AI is a cross-sectional technology with an almost universal scope of application, a particularly far-reaching effect of AI-related bias and discrimination on society is to be expected, which will spread virally in society especially via digital media. Given that AI is also still in its infancy with regard to research, development and use, it is therefore crucial to scrutinize its emerging risks and challenges as well as to supervise its development and dissemination in society (Weyerer & Langer, 2019).