ALGORITHMIC BIAS AS A SOCIOLOGICAL ISSUE: HOW BIG IS THE CHALLENGE FOR ‘AI FOR SOCIAL GOOD’?

Abstract

Artificial Intelligence (AI) undeniably has been the center of attention when it comes to technology and social change.  The ever- increasing penetration of AI applications to various segments of social and daily life is causing substantial positive and negative social impact.  One of the major reasons for the negative or unwanted social impact of AI is originating from AI bias rooted in the development process of AI itself, namely data collection and handling as well as the design of algorithms.  On the other hand, recent developments in AI and related technologies raised hope for positive social impact as well as finding solutions for the social and environmental problems of humanity. Artificial Intelligence for Social Good (AI4SG) concept has been developed with expectations to use state-of-the-art AI tools and applications for the benefit of whole society. However, AI bias is a significant challenge towards achieving the goals set by Artificial Intelligence for Social Good projects.  The alleviation of AI bias with reference to ethical and human-oriented guidelines may ensure successful results. In this regard, this paper discusses the challenges of AI toward achieving AI4SG goals.