“When Algorithms Harm: Consumers’ Responses following Algorithmic Bias”
by Gülen Sarıal Abi
Copenhagen Business School
Join Zoom Meeting
Meeting ID: 683 469 0808
Algorithms are widely used in marketing contexts including segmentation and targeting, product recommendations, pricing and advertising. Despite the efficiency gained through algorithmic technologies, there is increasing concern that algorithms may be biased, resulting in unfair outcomes for members of some protected classes. We examine how consumers respond to a brand that uses an algorithm following algorithmic bias resulting in algorithmic discrimination toward members of their protected class. Extending developments in the social categorization literature and its role in discrimination, we propose that consumers will respond negatively to a brand when the output of an algorithm used by the brand results in discrimination (vs. preferential treatment) toward members of their protected class. Consumers’ perceptions of the extent to which the algorithm uses social categorization that humans use in their decision-making will mediate the negative responses following bias that results in discrimination (vs. preferential treatment). We further propose that three sources of algorithmic bias (i.e., technical, emergent, and preexisting) will moderate consumers’ responses to the brand using the algorithm following algorithmic discrimination (vs. preferential treatment). The results from one secondary data study and five pre-registered experiments support the proposed theory and hypotheses. The findings generate insights on algorithmic bias and extend the literature on algorithmic marketing and also generate actionable guidelines for managerial practice.