Exploring the Ethical Implications of AI Algorithms in Decision-Making Systems

Abstract

Basic questions about transparency and equity in artificial intelligence (AI) arise when AI aids decision-making that impacts humans. There has been a lot of research on the links between AI-assisted decision-making and transparency/clarification and equity. Thus, considering their impact on customer trust or perceived fairness simultaneously aids in the thoughtful use of socio-specialized AI systems, which now receives less attention. In this study, we analyze fairness in AI-based decision-making scenarios that are free of ambiguity and the effects of AI explanations on human-AI trust. Significant experiences were gained from a customer research that replicated AI-assisted decision-making in two scenarios involving health care coverage and clinical treatment decision-making. Online overviews were used to lead the customer research because of the global epidemic and its restrictions. According to the members' trust perspective, clients' trust is affected by the amount of fairness, with a low level of fairness resulting in a decrease in customer trust. However, by providing more details, customers were able to increase their trust in decision-making supported by AI. Using the perspective of seen fairness, our research indicated that customers' perceptions of fairness were negatively impacted by low levels of presented fairness and positively impacted by high levels of presented fairness. Without a doubt, the perspective of fairness was broadened by the extension of explanations. Furthermore, we discovered that trust and perceptions of justice were impacted by application contexts. We need to think about not only the type of explanations and the level of fairness offered, but also the scenarios in which AI-aided decision-making is used, as the results demonstrate that the use of AI explanations and fairness proclamations in AI applications is really perplexing.