Responsible Machine Learning for Ethical Artificial Intelligence in Business and Industry

Introduction

Breaking away from the realm of science fiction, artificial intelligence (AI) systems have become the reality of modern life. Kaplan and Haenlein (2019) define AI as “a system’s ability to interpret external data correctly, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaptation”. AIs in modern life not only provide recommended course of actions to those working in business and industry, but they are also effectively making decisions for us. In our everyday life, they strongly influence what we watch during our leisure time, who we interact more with on social networks, the videos we watch and the music we listen to, what we buy on e-commerce websites and on what rate, what interest rate and credit limits we are offered on our credit cards, what route our taxis take and so on. The list of AI applications is endless. However, AI models are usually opaque, and one often knows very little on the decisions-making algorithm used by AIs (Burrell, 2016; de Saint Laurent, 2018; McQuillan, 2018; Mackenzie, 2019). Since AIs have so much influence on our lives, it is an imperative to focus our attention on the ethics of AI.

At the same time, AIs are built by humans, often using the process of machine learning. Hence, the choices made by the AI developers (who design the AI) and the managers (who decide to implement the recommendations from the AI) are crucial. This paper offers some suggestions towards responsible machine learning building on the idea of human-AI hybrid (Bansal et al., 2019a, 2019b; Dellermann et al., 2019; Jarrahi, 2018; Peeters et al., 2020). Taking a socio-technical systems perspective (Andras et al., 2018), the term human-AI hybrid indicates that each subsystem (AI and the humans) brings a unique skillset to the table and can help in ethical development of the overall system. In so doing, the chapter takes machine learning as a common denominator of AI and introduces a responsible machine learning framework towards ethical AI. For developers and managers, the chapter suggests actionable insights that can be used in industry and business.

The remainder of the chapter is as follows. The next section provides some background in terms of the notion of artificial intelligence, the machine learning process, and the concept of distributed morality in the context of socio-technical systems. Thereafter, the building blocks of responsible machine learning are discussed with the implications they have for the AI developers and managers. The building blocks are aligned with the classical machine learning stages. Based on the discussion, recommendations are presented for the managers and developers. This is followed by noting some limitations and future development, with last section concluding the chapter.