The Trust Paradox in AI-Driven Hiring: Socio-Cultural ‎Sensitivity As A Mediator of Fairness

Abstract

As Artificial Intelligence technology becomes part of recruitment processes, it is typically discussed as a remedy for human judgment and the handling of administrative tasks. However, we do not yet understand the social consequences of the use of these technologies at the pace at which they are accepted and used, which is creating a “trust paradox”. However, AI is being sold as objective, but often it reinforces inherent inequity and diminishes candidate trust. In this chapter, the problem that advanced algorithmic recruitment is not creating institutional or public trust is examined, and it's argued that technical approaches to ‘fairness’ shortcomings are not sufficient if they are not aligned with the varied, lived experiences of the global workforce. A Universal Metric is not necessarily appropriate for the local context. Standardized success measures, frequently tied to Western norms of workplace success, are typically used to train existing AI hiring systems on large datasets. Rather than this “universal candidate profile” - they basically try to do this - these models do not consider the fact that the meaning of “merit”, "potential" or “professionalism” is firmly rooted in cultural and socio-economic context. This section will take up the conflict between algorithmic uniformity and socio-cultural diversity. For example, the ‘gaps in employment’ or ‘non-traditional educational path’ disadvantages some data sets but doesn't for others where it is more a reflection of common cultural experience, such as precarious jobs or non-conventional educational journeys. First, it is not only a technical bias, but a failure to demonstrate cultural literacy in the design of the underlying AI models. These systems continue to reproduce inequalities, particularly by rejecting candidates who don't resemble the dominant culture in the training system, by defining “fit” as a rigid monolith. Theoretical Framework: Socio-Cultural Sensitivity as mediator This middle part asserts that to be understood as fair and, therefore, credible, AI should strive for a shift towards “socio-cultural sensitivity”. A framework that infuses socio-cultural processes as a necessary middleman in algorithmic decisions, using concepts from organizational behavior and cultural psychology, will be suggested. Cultural caliper of Merit: I will talk about how the AI architecture can be changed to embrace local conceptions of success. Systems should use multi-objective optimization of systems, based on the labor market constraints and values of the local market, instead of a single objective. This section will draw on my previous research into ‘green nudges’ and behavioral alignment to consider how trust is a result of the alignment between institutional values and candidate expectations. This perceived procedural justice arises as an AI tool displays an understanding of the cultural level that matches the candidate's community norms. The chapter will look at how socio-cultural sensitivity can be conveyed with Explainable AI (XAI) as a bridge. An ethically robust system should explain the decision in a culturally resonant way – meaning providing explanations for why a decision was made that also reflects the candidate's context in his or her locality at work. The world's implications and navigating the North-South Divide. Ethical issues surrounding AI in recruitment vary from country to country. In this section, it will be contrasted how AI is being used in markets when regulations are stringent (Global North) with those where regulations are emerging in the market (Global South). I will define hiring algorithms as a type of algorithmic colonialism if they are “exported” from one cultural context without adequate adaptation and thus imposed on another. I’ll argue that, when measuring the regulatory-equity gap, we can propose that TNC's should apply a strategy of “contextual compliance” whereby AI systems are notified for both mathematical neutrality and regulation according to the socio-cultural values and human rights of the area where they are used. The purpose of this chapter is to advance the Handbook of Business Ethics and Values in a Globalized World by moving beyond the dichotomy of “bias vs. fairness” to consider the issue of “contextual legitimacy”. In my contribution, I have two things I want to mention: Most of all, it provides a framework for HR practitioners to assess the “cultural fit” of AI platforms. Second, it creates a theoretical connecting “string” for scholars to walk across the digital-dialectic-technology divide, to grasp the fact that trust in technology is not only a technical issue but also a social one. I want to propose, in contrast, an alternative vision of an ethical HRM, in which the “human” of Human Resources is brought back, where local value systems are explicitly, actively, and responsibly incorporated into our future automated HR. In the end, AI-induced fairness is nothing more than a hollow charade, unless it can humble itself to acknowledge its cultural shortcomings in its logic.