Visions of Human-Centered Artificial Intelligence: Relations with Ethics and Power

Abstract

In recent years there has been heightened sensibility to ethical issues connected with artificial intelligence. Conceptions of different types of AI have emerged which touch upon this new sensibility, including wholesome augmented intelligence, responsible/trustworthy/ethical AI, AI for good, and human-centered AI. In this chapter, we focus on the concept of human-centered AI, which has gained more prominence lately and which has even appeared in the names of organizations. However, there are major problems with the conceptualization and operationalization of human-centered AI. This chapter critically analyzes academic visions about human-centered AI in five Western university institutions’ online textual content (n=573). The study scrutinizes institutions that use the term “human-centered” in their names. Even though institutions provide more content framed with supportive attitudes rather than focusing on technical solutions, the related texts fail to address several important issues. First, they often treat humanity as a homogenous group, suggesting that every society struggles with the same problems. Second, human-centered AI is treated as being mainly aligned with the Global North’s needs. Finally, most of the texts associated with the scrutinized institutions lack discussion of the surging inequalities connected to the capitalist system. Therefore, they do not offer many AI-supported solutions that might address the challenges of a lack of clean water, poverty, or the presence of dangerous jobs that harm the Global South. Instead, the analyzed institutions primarily present societal challenges within national borders, while they disregard the need for redressing fundamental problems that hinder the creation of acceptable living conditions in poor regions. Building on the sociology of expectations, this study argues that the visions of human-centered AI are of paramount importance. These expectations have the potential to legitimize, guide, and coordinate the activities of different actors responsible for the research, development, and application of AI-driven technologies.